release: v1.0.72

chore: format code
fix: noreply
2026-08-10 03:19:59 -04:00 · 2025-11-18 03:03:54 +00:00 · 2025-11-18 02:53:13 +00:00 · 2025-11-17 20:52:25 -06:00 · 2025-11-18 01:59:40 +00:00 · 2025-11-17 20:53:48 -05:00
92 changed files with 3795 additions and 2500 deletions
@@ -27,12 +27,12 @@ jobs:
            {
              "bash": {
                "gh issue*": "allow",
-                "*": "deny" 
-              }, 
+                "*": "deny"
+              },
              "webfetch": "deny"
            }
        run: |
-          opencode run -m anthropic/claude-sonnet-4-20250514 "A new issue has been created:'
+          opencode run -m opencode/claude-haiku-4-5 "A new issue has been created:'

          Issue number:
          ${{ github.event.issue.number }}
@@ -4,7 +4,7 @@ on:
  push:
    branches:
      - dev
-      - windows
+      - fix-build
      - v0

 concurrency: ${{ github.workflow }}-${{ github.ref }}
@@ -28,3 +28,9 @@ jobs:
          bun turbo test
        env:
          CI: true
+
+      - name: Check SDK is up to date
+        run: |
+          bun ./packages/sdk/js/script/build.ts
+          git diff --exit-code packages/sdk/js/src/gen packages/sdk/js/dist
+        continue-on-error: false
@@ -13,3 +13,6 @@ dist
 .turbo
 **/.serena
 .serena/
+refs
+Session.vim
+opencode.json
@@ -1,5 +1,6 @@
 ---
 description: Git commit and push
+subtask: true
 ---

 commit and push
@@ -0,0 +1,23 @@
+---
+description: "Find issue(s) on github"
+model: opencode/claude-haiku-4-5
+---
+
+Search through existing issues in sst/opencode using the gh cli to find issues matching this query:
+
+$ARGUMENTS
+
+Consider:
+
+1. Similar titles or descriptions
+2. Same error messages or symptoms
+3. Related functionality or components
+4. Similar feature requests
+
+Please list any matching issues with:
+
+- Issue number and title
+- Brief explanation of why it matches the query
+- Link to the issue
+
+If no clear matches are found, say so.
@@ -1,4 +0,0 @@
-{
-  "$schema": "https://opencode.ai/config.json",
-  "plugin": ["opencode-openai-codex-auth"]
-}
@@ -0,0 +1,11 @@
+{
+  "$schema": "https://opencode.ai/config.json",
+  "plugin": ["opencode-openai-codex-auth"],
+  "provider": {
+    "opencode": {
+      "options": {
+        // "baseURL": "http://localhost:8080",
+      },
+    },
+  },
+}
@@ -0,0 +1,11 @@
+{
+  "version": "0.2.0",
+  "configurations": [
+    {
+      "type": "bun",
+      "request": "attach",
+      "name": "opencode (attach)",
+      "url": "ws://localhost:6499/"
+    }
+  ]
+}
@@ -0,0 +1,5 @@
+{
+  "recommendations": [
+    "oven.bun-vscode"
+  ]
+}
@@ -42,6 +42,38 @@ Want to take on an issue? Leave a comment and a maintainer may assign it to you
 > [!NOTE]
 > After touching `packages/opencode/src/server/server.ts`, run "./packages/sdk/js/script/build.ts" to regenerate the JS sdk.

+### Setting up a Debugger
+
+Bun debugging is currently rough around the edges. We hope this guide helps you get set up and avoid some pain points.
+
+The most reliable way to debug OpenCode is to run it manually in a terminal via `bun run --inspect=<url> dev ...` and attach
+your debugger via that URL. Other methods can result in breakpoints being mapped incorrectly, at least in VSCode (YMMV).
+
+Caveats:
+
+- `*.tsx` files won't have their breakpoints correctly mapped. This seems due to Bun currently not supporting source maps on code transformed
+  via `BunPlugin`s (currently necessary due to our dependency on `@opentui/solid`). Currently, the best you can do in terms of debugging `*.tsx`
+  files is writing a `debugger;` statement. Debugging facilities like stepping won't work, but at least you will be informed if a specific code
+  is triggered.
+- If you want to run the OpenCode TUI and have breakpoints triggered in the server code, you might need to run `bun dev spawn` instead of
+  the usual `bun dev`. This is because `bun dev` runs the server in a worker thread and breakpoints might not work there.
+
+Other tips and tricks:
+
+- You might want to use `--inspect-wait` or `--inspect-brk` instead of `--inspect`, depending on your workflow
+- Specifying `--inspect=ws://localhost:6499/` on every invocation can be tiresome, you may want to `export BUN_OPTIONS=--inspect=ws://localhost:6499/` instead
+
+#### VSCode Setup
+
+If you use VSCode, you can use our example configurations [.vscode/settings.example.json](.vscode/settings.example.json) and [.vscode/launch.example.json](.vscode/launch.example.json).
+
+Some debug methods that can be problematic:
+
+- Debug configurations with `"request": "launch"` can have breakpoints incorrectly mapped and thus unusable
+- The same problem arises when running OpenCode in the VSCode `JavaScript Debug Terminal`
+
+With that said, you may want to try these methods, as they might work for you.
+
 ## Pull Request Expectations

 - Try to keep pull requests small and focused.
@@ -30,6 +30,7 @@ scoop bucket add extras; scoop install extras/opencode  # Windows
 choco install opencode             # Windows
 brew install opencode      # macOS and Linux
 paru -S opencode-bin               # Arch Linux
+mise use --pin -g ubi:sst/opencode # Any OS
 ```

 > [!TIP]
@@ -50,6 +51,22 @@ OPENCODE_INSTALL_DIR=/usr/local/bin curl -fsSL https://opencode.ai/install | bas
 XDG_BIN_DIR=$HOME/.local/bin curl -fsSL https://opencode.ai/install | bash
 ```

+### Agents
+
+OpenCode includes two built-in agents you can switch between,
+you can switch between these using the `Tab` key.
+
+- **build** - Default, full access agent for development work
+- **plan** - Read-only agent for analysis and code exploration
+  - Denies file edits by default
+  - Asks permission before running bash commands
+  - Ideal for exploring unfamiliar codebases or planning changes
+
+Also, included is a **general** subagent for complex searches and multi-step tasks.
+This is used internally and can be invoked using `@general` in messages.
+
+Learn more about [agents](https://opencode.ai/docs/agents).
+
 ### Documentation

 For more info on how to configure OpenCode [**head over to our docs**](https://opencode.ai/docs).
@@ -58,6 +75,10 @@ For more info on how to configure OpenCode [**head over to our docs**](https://o

 If you're interested in contributing to OpenCode, please read our [contributing docs](./CONTRIBUTING.md) before submitting a pull request.

+### Building on OpenCode
+
+If you are working on a project that's related to OpenCode and is using "opencode" as a part of its name; for example, "opencode-dashboard" or "opencode-mobile", please add a note to your README to clarify that it is not built by the OpenCode team and is not affiliated with us in anyway.
+
 ### FAQ

 #### How is this different than Claude Code?
@@ -139,3 +139,7 @@
 | 2025-11-11 | 729,769 (+7,481)  | 677,501 (+9,276)  | 1,407,270 (+16,757) |
 | 2025-11-12 | 740,180 (+10,411) | 686,454 (+8,953)  | 1,426,634 (+19,364) |
 | 2025-11-13 | 749,905 (+9,725)  | 696,157 (+9,703)  | 1,446,062 (+19,428) |
+| 2025-11-14 | 759,928 (+10,023) | 705,237 (+9,080)  | 1,465,165 (+19,103) |
+| 2025-11-15 | 765,955 (+6,027)  | 712,870 (+7,633)  | 1,478,825 (+13,660) |
+| 2025-11-16 | 771,069 (+5,114)  | 716,596 (+3,726)  | 1,487,665 (+8,840)  |
+| 2025-11-17 | 780,161 (+9,092)  | 723,339 (+6,743)  | 1,503,500 (+15,835) |
@@ -40,7 +40,7 @@
    },
    "packages/console/core": {
      "name": "@opencode-ai/console-core",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@aws-sdk/client-sts": "3.782.0",
        "@jsx-email/render": "1.1.1",
@@ -67,7 +67,7 @@
    },
    "packages/console/function": {
      "name": "@opencode-ai/console-function",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@ai-sdk/anthropic": "2.0.0",
        "@ai-sdk/openai": "2.0.2",
@@ -91,7 +91,7 @@
    },
    "packages/console/mail": {
      "name": "@opencode-ai/console-mail",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@jsx-email/all": "2.2.3",
        "@jsx-email/cli": "1.4.3",
@@ -115,7 +115,7 @@
    },
    "packages/desktop": {
      "name": "@opencode-ai/desktop",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@kobalte/core": "catalog:",
        "@opencode-ai/sdk": "workspace:*",
@@ -155,7 +155,7 @@
    },
    "packages/function": {
      "name": "@opencode-ai/function",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@octokit/auth-app": "8.0.1",
        "@octokit/rest": "22.0.0",
@@ -171,7 +171,7 @@
    },
    "packages/opencode": {
      "name": "opencode",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "bin": {
        "opencode": "./bin/opencode",
      },
@@ -189,8 +189,8 @@
        "@opencode-ai/plugin": "workspace:*",
        "@opencode-ai/script": "workspace:*",
        "@opencode-ai/sdk": "workspace:*",
-        "@opentui/core": "0.1.42",
-        "@opentui/solid": "0.1.42",
+        "@opentui/core": "0.1.46",
+        "@opentui/solid": "0.1.46",
        "@parcel/watcher": "2.5.1",
        "@pierre/precision-diffs": "catalog:",
        "@solid-primitives/event-bus": "1.1.2",
@@ -249,7 +249,7 @@
    },
    "packages/plugin": {
      "name": "@opencode-ai/plugin",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@opencode-ai/sdk": "workspace:*",
        "zod": "catalog:",
@@ -269,7 +269,7 @@
    },
    "packages/sdk/js": {
      "name": "@opencode-ai/sdk",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "devDependencies": {
        "@hey-api/openapi-ts": "0.81.0",
        "@tsconfig/node22": "catalog:",
@@ -280,7 +280,7 @@
    },
    "packages/slack": {
      "name": "@opencode-ai/slack",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@opencode-ai/sdk": "workspace:*",
        "@slack/bolt": "^3.17.1",
@@ -293,7 +293,7 @@
    },
    "packages/ui": {
      "name": "@opencode-ai/ui",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@kobalte/core": "catalog:",
        "@opencode-ai/sdk": "workspace:*",
@@ -323,7 +323,7 @@
    },
    "packages/web": {
      "name": "@opencode-ai/web",
-      "version": "1.0.64",
+      "version": "1.0.72",
      "dependencies": {
        "@astrojs/cloudflare": "12.6.3",
        "@astrojs/markdown-remark": "6.3.1",
@@ -966,21 +966,21 @@

    "@opentelemetry/api": ["@opentelemetry/api@1.9.0", "", {}, "sha512-3giAOQvZiH5F9bMlMiv8+GSPMeqg0dbaeo58/0SlA9sxSqZhnUtxzX9/2FzyhS9sWQf5S0GJE0AKBrFqjpeYcg=="],

-    "@opentui/core": ["@opentui/core@0.1.42", "", { "dependencies": { "bun-ffi-structs": "0.1.2", "jimp": "1.6.0", "yoga-layout": "3.2.1" }, "optionalDependencies": { "@dimforge/rapier2d-simd-compat": "^0.17.3", "@opentui/core-darwin-arm64": "0.1.42", "@opentui/core-darwin-x64": "0.1.42", "@opentui/core-linux-arm64": "0.1.42", "@opentui/core-linux-x64": "0.1.42", "@opentui/core-win32-arm64": "0.1.42", "@opentui/core-win32-x64": "0.1.42", "bun-webgpu": "0.1.3", "planck": "^1.4.2", "three": "0.177.0" }, "peerDependencies": { "web-tree-sitter": "0.25.10" } }, "sha512-oV2xHBB2HaNiGvaV6R0C8GmniNJSsLKop4APq4FrLyCYberc6vZcATSHcA5YT9krdvHbBDOOn9RI2oaVJYRbUQ=="],
+    "@opentui/core": ["@opentui/core@0.1.46", "", { "dependencies": { "bun-ffi-structs": "0.1.2", "jimp": "1.6.0", "yoga-layout": "3.2.1" }, "optionalDependencies": { "@dimforge/rapier2d-simd-compat": "^0.17.3", "@opentui/core-darwin-arm64": "0.1.46", "@opentui/core-darwin-x64": "0.1.46", "@opentui/core-linux-arm64": "0.1.46", "@opentui/core-linux-x64": "0.1.46", "@opentui/core-win32-arm64": "0.1.46", "@opentui/core-win32-x64": "0.1.46", "bun-webgpu": "0.1.4", "planck": "^1.4.2", "three": "0.177.0" }, "peerDependencies": { "web-tree-sitter": "0.25.10" } }, "sha512-9682jrr65vYP0hPHfrZRK3xymlCSLVBMrRKNtclFasDi6bRvACUrtziFOIIyMIvPHRJCFWPbtz0MppARmN4zvQ=="],

-    "@opentui/core-darwin-arm64": ["@opentui/core-darwin-arm64@0.1.42", "", { "os": "darwin", "cpu": "arm64" }, "sha512-Sk5b/kh/y8HUJ7stGA5ydkajJX/z2OiGqSm+wn6XIoqdDavxQaFoQOt1PCuCqaxqZWJcXZ6OmISDVagZPUsPuw=="],
+    "@opentui/core-darwin-arm64": ["@opentui/core-darwin-arm64@0.1.46", "", { "os": "darwin", "cpu": "arm64" }, "sha512-Sp/uFS8J/1yVfhgkTJ43OZfy64hv1/9sdT+oC5yb8XTNPI3QGtg6ixjr3nRoD/Lkxuj2i5SJ30RZufqH6rkCpA=="],

-    "@opentui/core-darwin-x64": ["@opentui/core-darwin-x64@0.1.42", "", { "os": "darwin", "cpu": "x64" }, "sha512-b0FKTw+t/wlJg4u+wTurWzbQe47gExkjguaGSUua0m0vybrkkvbUvmrADr+yivCjxcPAhSZ3lOOVU3uZuWsNqw=="],
+    "@opentui/core-darwin-x64": ["@opentui/core-darwin-x64@0.1.46", "", { "os": "darwin", "cpu": "x64" }, "sha512-JtxEruRyLQRK8ByPmBm1nYYSvnX6mXNC+mngvd5RDiCzLzkM6qVBQBd/m3Hxp2/s6MO5Z2+iVBzZ8XFH5T4IZw=="],

-    "@opentui/core-linux-arm64": ["@opentui/core-linux-arm64@0.1.42", "", { "os": "linux", "cpu": "arm64" }, "sha512-Vy8BrjJpv2f56JAsYmv4PkC+2HsCv8Gh0ErrlIJQ8L4h29oWabS44m0uxFdvjuTDgKpCJzOScsxsy1VGzSd9rw=="],
+    "@opentui/core-linux-arm64": ["@opentui/core-linux-arm64@0.1.46", "", { "os": "linux", "cpu": "arm64" }, "sha512-pN8nR4CwBlkZ5uh5KvoytiXXav2GhkP9cB2d3gPe49c7MBz2XrjGexgb47xjaq0hAVbytv9XUifqdPTcFQdPaQ=="],

-    "@opentui/core-linux-x64": ["@opentui/core-linux-x64@0.1.42", "", { "os": "linux", "cpu": "x64" }, "sha512-cO+13E1HIAPUdV/DRdKotHFAxsLc+ipbbFKGAuu/msfvywCnnNs86w22yeMg0cEqx7aBocWWT1XfJEHDJLFOqw=="],
+    "@opentui/core-linux-x64": ["@opentui/core-linux-x64@0.1.46", "", { "os": "linux", "cpu": "x64" }, "sha512-oH4/FEYZYce9qMQVqGl4+Btw+Mfsf6ybpWIIJUJjXMWWZlAgsTMAWM8m195Oe6WstfFLF+nRH7NUcm/YOsCHnw=="],

-    "@opentui/core-win32-arm64": ["@opentui/core-win32-arm64@0.1.42", "", { "os": "win32", "cpu": "arm64" }, "sha512-xpLhODjOWh7gMOSrKIldb4v6hR0TGyz6kjckDKwcjUv3LGbLJuSly+3O/zuWWS60dt56G1X4A0OyjWwiGZjc0g=="],
+    "@opentui/core-win32-arm64": ["@opentui/core-win32-arm64@0.1.46", "", { "os": "win32", "cpu": "arm64" }, "sha512-C/rTBJ9bzBcZJRCLIxi9Ka/DANe2SaHtryotseWPk9RDydw7LTHGoi3VtRW0RFijQGqmvFg+31MeNhvY1YZ65Q=="],

-    "@opentui/core-win32-x64": ["@opentui/core-win32-x64@0.1.42", "", { "os": "win32", "cpu": "x64" }, "sha512-pao5XdAln93WWPdsTF+V+HccZ5d1ijSmv0OoBbkjkVbP+tiN41yxNqg/7jzW9IiAakYsvmpKV+3ixi/dlBEvOQ=="],
+    "@opentui/core-win32-x64": ["@opentui/core-win32-x64@0.1.46", "", { "os": "win32", "cpu": "x64" }, "sha512-d2DXSlA93LbSriX+pDDZ5sMwkcW1+eVoeykxeW4UParSb4/3ceBCD4aSARaZ6yoq0rR1IWOdgKdiihZH4mwdJQ=="],

-    "@opentui/solid": ["@opentui/solid@0.1.42", "", { "dependencies": { "@babel/core": "7.28.0", "@babel/preset-typescript": "7.27.1", "@opentui/core": "0.1.42", "babel-plugin-module-resolver": "5.0.2", "babel-preset-solid": "1.9.9", "s-js": "^0.4.9" }, "peerDependencies": { "solid-js": "1.9.9" } }, "sha512-4TNlEtatZ4n9TcKPWSF/EoaPaLmZuFVJ4hHh9wRggNaGrmDlmJ+9N/8oEKXETt+oRDX/1CdowAaTOVfaqb1t6g=="],
+    "@opentui/solid": ["@opentui/solid@0.1.46", "", { "dependencies": { "@babel/core": "7.28.0", "@babel/preset-typescript": "7.27.1", "@opentui/core": "0.1.46", "babel-plugin-module-resolver": "5.0.2", "babel-preset-solid": "1.9.9", "s-js": "^0.4.9" }, "peerDependencies": { "solid-js": "1.9.9" } }, "sha512-t+LDXS+FBT+fjQHOxuL44Bx1jGuXuLouyB25BZuypQKKT8sOhi8rJ5+Q4UwH5lrI4OoRBXmzrgsWj7DD58sHDw=="],

    "@oslojs/asn1": ["@oslojs/asn1@1.0.0", "", { "dependencies": { "@oslojs/binary": "1.0.0" } }, "sha512-zw/wn0sj0j0QKbIXfIlnEcTviaCzYOY3V5rAyjR6YtOByFtJiT574+8p9Wlach0lZH9fddD4yb9laEAIl4vXQA=="],

@@ -1684,15 +1684,15 @@

    "bun-types": ["bun-types@1.3.0", "", { "dependencies": { "@types/node": "*" }, "peerDependencies": { "@types/react": "^19" } }, "sha512-u8X0thhx+yJ0KmkxuEo9HAtdfgCBaM/aI9K90VQcQioAmkVp3SG3FkwWGibUFz3WdXAdcsqOcbU40lK7tbHdkQ=="],

-    "bun-webgpu": ["bun-webgpu@0.1.3", "", { "dependencies": { "@webgpu/types": "^0.1.60" }, "optionalDependencies": { "bun-webgpu-darwin-arm64": "^0.1.3", "bun-webgpu-darwin-x64": "^0.1.3", "bun-webgpu-linux-x64": "^0.1.3", "bun-webgpu-win32-x64": "^0.1.3" } }, "sha512-IXFxaIi4rgsEEpl9n/QVDm5RajCK/0FcOXZeMb52YRjoiAR1YVYK5hLrXT8cm+KDi6LVahA9GJFqOR4yiloVCw=="],
+    "bun-webgpu": ["bun-webgpu@0.1.4", "", { "dependencies": { "@webgpu/types": "^0.1.60" }, "optionalDependencies": { "bun-webgpu-darwin-arm64": "^0.1.4", "bun-webgpu-darwin-x64": "^0.1.4", "bun-webgpu-linux-x64": "^0.1.4", "bun-webgpu-win32-x64": "^0.1.4" } }, "sha512-Kw+HoXl1PMWJTh9wvh63SSRofTA8vYBFCw0XEP1V1fFdQEDhI8Sgf73sdndE/oDpN/7CMx0Yv/q8FCvO39ROMQ=="],

-    "bun-webgpu-darwin-arm64": ["bun-webgpu-darwin-arm64@0.1.3", "", { "os": "darwin", "cpu": "arm64" }, "sha512-KkNQ9gT7dxGDndQaHTTHss9miukqpczML3pO2nZJoT/nITwe9lw3ZGFJMujkW41BUQ1mDYKFgo5nBGf9xYHPAg=="],
+    "bun-webgpu-darwin-arm64": ["bun-webgpu-darwin-arm64@0.1.4", "", { "os": "darwin", "cpu": "arm64" }, "sha512-eDgLN9teKTfmvrCqgwwmWNsNszxYs7IZdCqk0S1DCarvMhr4wcajoSBlA/nQA0/owwLduPTS8xxCnQp4/N/gDg=="],

-    "bun-webgpu-darwin-x64": ["bun-webgpu-darwin-x64@0.1.3", "", { "os": "darwin", "cpu": "x64" }, "sha512-TODWnMUbCoqD/wqzlB3oGOBIUWIFly0lqMeBFz/MBV+ndjbnkNrP9huaZJCTkCVEPKGtd1FCM3ExZUtBbnGziA=="],
+    "bun-webgpu-darwin-x64": ["bun-webgpu-darwin-x64@0.1.4", "", { "os": "darwin", "cpu": "x64" }, "sha512-X+PjwJUWenUmdQBP8EtdItMyieQ6Nlpn+BH518oaouDiSnWj5+b0Y7DNDZJq7Ezom4EaxmqL/uGYZK3aCQ7CXg=="],

-    "bun-webgpu-linux-x64": ["bun-webgpu-linux-x64@0.1.3", "", { "os": "linux", "cpu": "x64" }, "sha512-lVHORoVu1G61XVM8CRRqUsqr6w8kMlpuSpbPGpKUpmvrsoay6ymXAhT5lRPKyrGNamHUQTknmWdI59aRDCfLtQ=="],
+    "bun-webgpu-linux-x64": ["bun-webgpu-linux-x64@0.1.4", "", { "os": "linux", "cpu": "x64" }, "sha512-zMLs2YIGB+/jxrYFXaFhVKX/GBt05UTF45lc9srcHc9JXGjEj+12CIo1CHLTAWatXMTqt0Jsu6ukWEoWVT/ayA=="],

-    "bun-webgpu-win32-x64": ["bun-webgpu-win32-x64@0.1.3", "", { "os": "win32", "cpu": "x64" }, "sha512-vlspsFffctJlBnFfs2lW3QgDD6LyFu8VT18ryID7Qka5poTj0clGVRxz7DFRi7yva3GovEGw/82z/WVc5US8Pw=="],
+    "bun-webgpu-win32-x64": ["bun-webgpu-win32-x64@0.1.4", "", { "os": "win32", "cpu": "x64" }, "sha512-Z5yAK28xrcm8Wb5k7TZ8FJKpOI/r+aVCRdlHYAqI2SDJFN3nD4mJs900X6kNVmG/xFzb5yOuKVYWGg+6ZXWbyA=="],

    "bundle-name": ["bundle-name@4.1.0", "", { "dependencies": { "run-applescript": "^7.0.0" } }, "sha512-tjwM5exMg6BGRI+kNmTntNsvdZS1X8BFYS6tnJ2hdH0kVxM6/eVZ2xy+FqStSWvYmtfFMDLIxurorHwDKfDz5Q=="],

@@ -7,7 +7,7 @@
    "dev:remote": "VITE_AUTH_URL=https://auth.dev.opencode.ai bun sst shell --stage=dev bun dev",
    "build": "./script/generate-sitemap.ts && vinxi build && ../../opencode/script/schema.ts ./.output/public/config.json",
    "start": "vinxi start",
-    "version": "1.0.64"
+    "version": "1.0.72"
  },
  "dependencies": {
    "@ibm/plex": "6.4.1",
@@ -22,8 +22,8 @@ const getModelsInfo = query(async (workspaceID: string) => {
  return withActor(async () => {
    return {
      all: Object.entries(ZenData.list().models)
-        .filter(([id, _model]) => !["claude-3-5-haiku", "minimax-m2"].includes(id))
-        .filter(([id, _model]) => !id.startsWith("an-"))
+        .filter(([id, _model]) => !["claude-3-5-haiku"].includes(id))
+        .filter(([id, _model]) => !id.startsWith("alpha-"))
        .sort(([_idA, modelA], [_idB, modelB]) => modelA.name.localeCompare(modelB.name))
        .map(([id, model]) => ({ id, name: model.name })),
      disabled: await Model.listDisabled(),
@@ -1,88 +1,111 @@
-.root {
-  [data-component="empty-state"] {
-    padding: var(--space-20) var(--space-6);
-    text-align: center;
-    border: 1px dashed var(--color-border);
-    border-radius: var(--border-radius-sm);
-    display: flex;
-    flex-direction: column;
-    gap: var(--space-2);
+/* Empty state */
+[data-component="empty-state"] {
+  padding: var(--space-20) var(--space-6);
+  text-align: center;
+  border: 1px dashed var(--color-border);
+  border-radius: var(--border-radius-sm);

-    p {
-      line-height: 1.5;
-      font-size: var(--font-size-sm);
-      color: var(--color-text-muted);
-    }
-  }
-
-  [data-slot="usage-table"] {
-    overflow-x: auto;
-  }
-
-  [data-slot="usage-table-element"] {
-    width: 100%;
-    border-collapse: collapse;
+  p {
    font-size: var(--font-size-sm);
+    color: var(--color-text-muted);
+  }
+}

-    thead {
-      border-bottom: 1px solid var(--color-border);
+/* Table container */
+[data-slot="usage-table"] {
+  overflow-x: auto;
+}
+
+/* Table element */
+[data-slot="usage-table-element"] {
+  width: 100%;
+  border-collapse: collapse;
+  font-size: var(--font-size-sm);
+
+  thead {
+    border-bottom: 1px solid var(--color-border);
+  }
+
+  th {
+    padding: var(--space-3) var(--space-4);
+    text-align: left;
+    font-weight: normal;
+    color: var(--color-text-muted);
+    text-transform: uppercase;
+  }
+
+  td {
+    padding: var(--space-3) var(--space-4);
+    border-bottom: 1px solid var(--color-border-muted);
+    color: var(--color-text-muted);
+    font-family: var(--font-mono);
+
+    &[data-slot="usage-date"] {
+      color: var(--color-text);
    }

-    th {
-      padding: var(--space-3) var(--space-4);
-      text-align: left;
-      font-weight: normal;
-      color: var(--color-text-muted);
-      text-transform: uppercase;
+    &[data-slot="usage-model"] {
+      font-family: var(--font-sans);
+      color: var(--color-text-secondary);
+      max-width: 200px;
+      word-break: break-word;
    }

-    td {
-      padding: var(--space-3) var(--space-4);
-      border-bottom: 1px solid var(--color-border-muted);
-      color: var(--color-text-muted);
-      font-family: var(--font-mono);
+    &[data-slot="usage-cost"] {
+      color: var(--color-text);
+      font-weight: 500;
+    }
+  }

-      &[data-slot="usage-date"] {
-        color: var(--color-text);
-      }
+  tbody tr:last-child td {
+    border-bottom: none;
+  }
+}

-      &[data-slot="usage-model"] {
-        font-family: var(--font-sans);
-        font-weight: 400;
-        color: var(--color-text-secondary);
-        max-width: 200px;
-        word-break: break-word;
-      }
+/* Pagination */
+[data-slot="pagination"] {
+  display: flex;
+  justify-content: flex-end;
+  gap: var(--space-2);
+  padding: var(--space-4) 0;
+  border-top: 1px solid var(--color-border-muted);
+  margin-top: var(--space-2);

-      &[data-slot="usage-cost"] {
-        color: var(--color-text);
-      }
+  button {
+    padding: var(--space-2) var(--space-4);
+    background: var(--color-bg-secondary);
+    border: 1px solid var(--color-border);
+    border-radius: var(--border-radius-sm);
+    color: var(--color-text);
+    font-size: var(--font-size-sm);
+    cursor: pointer;
+    transition: all 0.15s ease;
+
+    &:hover:not(:disabled) {
+      background: var(--color-bg-tertiary);
+      border-color: var(--color-border-hover);
    }

-    tbody tr {
-      &:last-child td {
-        border-bottom: none;
-      }
-    }
-
-    @media (max-width: 40rem) {
-      th,
-      td {
-        padding: var(--space-2) var(--space-3);
-        font-size: var(--font-size-xs);
-      }
-
-      th {
-        &:nth-child(2) /* Model */ {
-          display: none;
-        }
-      }
-
-      td {
-        &:nth-child(2) /* Model */ {
-          display: none;
-        }
-      }
+    &:disabled {
+      opacity: 0.5;
+      cursor: not-allowed;
+    }
+  }
+}
+
+/* Mobile responsive */
+@media (max-width: 40rem) {
+  [data-slot="usage-table-element"] {
+    th,
+    td {
+      padding: var(--space-2) var(--space-3);
+      font-size: var(--font-size-xs);
+    }
+
+    /* Hide Model column on mobile */
+    th:nth-child(2),
+    td:nth-child(2) {
+      display: none;
    }
  }
 }
@@ -1,91 +1,59 @@
 import { Billing } from "@opencode-ai/console-core/billing.js"
-import { query, useParams, createAsync } from "@solidjs/router"
-import { createMemo, For, Show } from "solid-js"
+import { createAsync, query, useParams } from "@solidjs/router"
+import { createMemo, For, Show, createEffect } from "solid-js"
 import { formatDateUTC, formatDateForTable } from "../common"
 import { withActor } from "~/context/auth.withActor"
-import styles from "./usage-section.module.css"
+import "./usage-section.module.css"
+import { createStore } from "solid-js/store"

-const getUsageInfo = query(async (workspaceID: string) => {
+const PAGE_SIZE = 50
+
+async function getUsageInfo(workspaceID: string, page: number) {
  "use server"
  return withActor(async () => {
-    return await Billing.usages()
+    return await Billing.usages(page, PAGE_SIZE)
  }, workspaceID)
-}, "usage.list")
+}
+
+const queryUsageInfo = query(getUsageInfo, "usage.list")

 export function UsageSection() {
  const params = useParams()
-  // ORIGINAL CODE - COMMENTED OUT FOR TESTING
-  const usage = createAsync(() => getUsageInfo(params.id!))
+  const usage = createAsync(() => queryUsageInfo(params.id!, 0))
+  const [store, setStore] = createStore({ page: 0, usage: [] as Awaited<ReturnType<typeof getUsageInfo>> })

-  // DUMMY DATA FOR TESTING
-  // const usage = () => [
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 0).toISOString(), // Today
-  //     model: "claude-3-5-sonnet-20241022",
-  //     inputTokens: 1247,
-  //     outputTokens: 423,
-  //     cost: 125400000, // $1.254
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 0.5).toISOString(), // 12 hours ago
-  //     model: "claude-3-haiku-20240307",
-  //     inputTokens: 892,
-  //     outputTokens: 156,
-  //     cost: 23500000, // $0.235
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 1).toISOString(), // Yesterday
-  //     model: "claude-3-5-sonnet-20241022",
-  //     inputTokens: 2134,
-  //     outputTokens: 687,
-  //     cost: 234700000, // $2.347
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 1.3).toISOString(), // 1.3 days ago
-  //     model: "gpt-4o-mini",
-  //     inputTokens: 567,
-  //     outputTokens: 234,
-  //     cost: 8900000, // $0.089
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 2).toISOString(), // 2 days ago
-  //     model: "claude-3-opus-20240229",
-  //     inputTokens: 1893,
-  //     outputTokens: 945,
-  //     cost: 445600000, // $4.456
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 2.7).toISOString(), // 2.7 days ago
-  //     model: "gpt-4o",
-  //     inputTokens: 1456,
-  //     outputTokens: 532,
-  //     cost: 156800000, // $1.568
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 3).toISOString(), // 3 days ago
-  //     model: "claude-3-haiku-20240307",
-  //     inputTokens: 634,
-  //     outputTokens: 89,
-  //     cost: 12300000, // $0.123
-  //   },
-  //   {
-  //     timeCreated: new Date(Date.now() - 86400000 * 4).toISOString(), // 4 days ago
-  //     model: "claude-3-5-sonnet-20241022",
-  //     inputTokens: 3245,
-  //     outputTokens: 1123,
-  //     cost: 387200000, // $3.872
-  //   },
-  // ]
+  createEffect(() => {
+    setStore({ usage: usage() })
+  }, [usage])
+
+  const hasResults = createMemo(() => store.usage && store.usage.length > 0)
+  const canGoPrev = createMemo(() => store.page > 0)
+  const canGoNext = createMemo(() => store.usage && store.usage.length === PAGE_SIZE)
+
+  const goPrev = async () => {
+    const usage = await getUsageInfo(params.id!, store.page - 1)
+    setStore({
+      page: store.page - 1,
+      usage,
+    })
+  }
+  const goNext = async () => {
+    const usage = await getUsageInfo(params.id!, store.page + 1)
+    setStore({
+      page: store.page + 1,
+      usage,
+    })
+  }

  return (
-    <section class={styles.root}>
+    <section>
      <div data-slot="section-title">
        <h2>Usage History</h2>
        <p>Recent API usage and costs.</p>
      </div>
      <div data-slot="usage-table">
        <Show
-          when={usage() && usage()!.length > 0}
+          when={hasResults()}
          fallback={
            <div data-component="empty-state">
              <p>Make your first API call to get started.</p>
@@ -103,7 +71,7 @@ export function UsageSection() {
              </tr>
            </thead>
            <tbody>
-              <For each={usage()!}>
+              <For each={store.usage}>
                {(usage) => {
                  const date = createMemo(() => new Date(usage.timeCreated))
                  return (
@@ -121,6 +89,16 @@ export function UsageSection() {
              </For>
            </tbody>
          </table>
+          <Show when={canGoPrev() || canGoNext()}>
+            <div data-slot="pagination">
+              <button disabled={!canGoPrev()} onClick={goPrev}>
+                ←
+              </button>
+              <button disabled={!canGoNext()} onClick={goNext}>
+                →
+              </button>
+            </div>
+          </Show>
        </Show>
      </div>
    </section>
@@ -291,7 +291,7 @@ export async function handler(

  async function authenticate(modelInfo: ModelInfo, providerInfo: ProviderInfo) {
    const apiKey = opts.parseApiKey(input.request.headers)
-    if (!apiKey) {
+    if (!apiKey || apiKey === "public") {
      if (modelInfo.allowAnonymous) return
      throw new AuthError("Missing API key.")
    }
@@ -1,7 +1,7 @@
 {
  "$schema": "https://json.schemastore.org/package.json",
  "name": "@opencode-ai/console-core",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "private": true,
  "type": "module",
  "dependencies": {
@@ -57,14 +57,15 @@ export namespace Billing {
    )
  }

-  export const usages = async () => {
+  export const usages = async (page = 0, pageSize = 50) => {
    return await Database.use((tx) =>
      tx
        .select()
        .from(UsageTable)
        .where(eq(UsageTable.workspaceID, Actor.workspace()))
        .orderBy(sql`${UsageTable.timeCreated} DESC`)
-        .limit(100),
+        .limit(pageSize)
+        .offset(page * pageSize),
    )
  }

@@ -1,6 +1,6 @@
 {
  "name": "@opencode-ai/console-function",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "$schema": "https://json.schemastore.org/package.json",
  "private": true,
  "type": "module",
@@ -1,6 +1,6 @@
 {
  "name": "@opencode-ai/console-mail",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "dependencies": {
    "@jsx-email/all": "2.2.3",
    "@jsx-email/cli": "1.4.3",
@@ -1,6 +1,6 @@
 {
  "name": "@opencode-ai/desktop",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "description": "",
  "type": "module",
  "scripts": {
@@ -1,7 +1,7 @@
 id = "opencode"
 name = "OpenCode"
 description = "The AI coding agent built for the terminal"
-version = "1.0.64"
+version = "1.0.72"
 schema_version = 1
 authors = ["Anomaly"]
 repository = "https://github.com/sst/opencode"
@@ -11,26 +11,26 @@ name = "OpenCode"
 icon = "./icons/opencode.svg"

 [agent_servers.opencode.targets.darwin-aarch64]
-archive = "https://github.com/sst/opencode/releases/download/v1.0.64/opencode-darwin-arm64.zip"
+archive = "https://github.com/sst/opencode/releases/download/v1.0.72/opencode-darwin-arm64.zip"
 cmd = "./opencode"
 args = ["acp"]

 [agent_servers.opencode.targets.darwin-x86_64]
-archive = "https://github.com/sst/opencode/releases/download/v1.0.64/opencode-darwin-x64.zip"
+archive = "https://github.com/sst/opencode/releases/download/v1.0.72/opencode-darwin-x64.zip"
 cmd = "./opencode"
 args = ["acp"]

 [agent_servers.opencode.targets.linux-aarch64]
-archive = "https://github.com/sst/opencode/releases/download/v1.0.64/opencode-linux-arm64.zip"
+archive = "https://github.com/sst/opencode/releases/download/v1.0.72/opencode-linux-arm64.zip"
 cmd = "./opencode"
 args = ["acp"]

 [agent_servers.opencode.targets.linux-x86_64]
-archive = "https://github.com/sst/opencode/releases/download/v1.0.64/opencode-linux-x64.zip"
+archive = "https://github.com/sst/opencode/releases/download/v1.0.72/opencode-linux-x64.zip"
 cmd = "./opencode"
 args = ["acp"]

 [agent_servers.opencode.targets.windows-x86_64]
-archive = "https://github.com/sst/opencode/releases/download/v1.0.64/opencode-windows-x64.zip"
+archive = "https://github.com/sst/opencode/releases/download/v1.0.72/opencode-windows-x64.zip"
 cmd = "./opencode.exe"
 args = ["acp"]
@@ -1,6 +1,6 @@
 {
  "name": "@opencode-ai/function",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "$schema": "https://json.schemastore.org/package.json",
  "private": true,
  "type": "module",
@@ -1,6 +1,6 @@
 {
  "$schema": "https://json.schemastore.org/package.json",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "name": "opencode",
  "type": "module",
  "private": true,
@@ -54,8 +54,8 @@
    "@opencode-ai/plugin": "workspace:*",
    "@opencode-ai/script": "workspace:*",
    "@opencode-ai/sdk": "workspace:*",
-    "@opentui/core": "0.1.42",
-    "@opentui/solid": "0.1.42",
+    "@opentui/core": "0.1.46",
+    "@opentui/solid": "0.1.46",
    "@parcel/watcher": "2.5.1",
    "@pierre/precision-diffs": "catalog:",
    "@solid-primitives/event-bus": "1.1.2",
@@ -1,6 +1,5 @@
 #!/usr/bin/env bun

-import solidPlugin from "../node_modules/@opentui/solid/scripts/solid-plugin"
 import path from "path"
 import fs from "fs"
 import { $ } from "bun"
@@ -10,6 +9,9 @@ const __filename = fileURLToPath(import.meta.url)
 const __dirname = path.dirname(__filename)
 const dir = path.resolve(__dirname, "..")

+const solidPluginPath = path.resolve(dir, "node_modules/@opentui/solid/scripts/solid-plugin.ts")
+const solidPlugin = (await import(solidPluginPath)).default
+
 process.chdir(dir)

 import pkg from "../package.json"
@@ -12,7 +12,7 @@ export namespace Agent {
    .object({
      name: z.string(),
      description: z.string().optional(),
-      mode: z.union([z.literal("subagent"), z.literal("primary"), z.literal("all")]),
+      mode: z.enum(["subagent", "primary", "all"]),
      builtIn: z.boolean(),
      topP: z.number().optional(),
      temperature: z.number().optional(),
@@ -0,0 +1,10 @@
+import { EventEmitter } from "events"
+
+export const GlobalBus = new EventEmitter<{
+  event: [
+    {
+      directory: string
+      payload: any
+    },
+  ]
+}>()
@@ -2,6 +2,7 @@ import z from "zod"
 import type { ZodType } from "zod"
 import { Log } from "../util/log"
 import { Instance } from "../project/instance"
+import { GlobalBus } from "./global"

 export namespace Bus {
  const log = Log.create({ service: "bus" })
@@ -29,22 +30,26 @@ export namespace Bus {
  }

  export function payloads() {
-    return z.discriminatedUnion(
-      "type",
-      registry
-        .entries()
-        .map(([type, def]) => {
-          return z
-            .object({
-              type: z.literal(type),
-              properties: def.properties,
-            })
-            .meta({
-              ref: "Event" + "." + def.type,
-            })
-        })
-        .toArray() as any,
-    )
+    return z
+      .discriminatedUnion(
+        "type",
+        registry
+          .entries()
+          .map(([type, def]) => {
+            return z
+              .object({
+                type: z.literal(type),
+                properties: def.properties,
+              })
+              .meta({
+                ref: "Event" + "." + def.type,
+              })
+          })
+          .toArray() as any,
+      )
+      .meta({
+        ref: "Event",
+      })
  }

  export async function publish<Definition extends EventDefinition>(
@@ -65,6 +70,10 @@ export namespace Bus {
        pending.push(sub(payload))
      }
    }
+    GlobalBus.emit("event", {
+      directory: Instance.directory,
+      payload,
+    })
    return Promise.all(pending)
  }

@@ -2,6 +2,7 @@ import { EOL } from "os"
 import { File } from "../../../file"
 import { bootstrap } from "../../bootstrap"
 import { cmd } from "../cmd"
+import { Ripgrep } from "@/file/ripgrep"

 const FileSearchCommand = cmd({
  command: "search <query>",
@@ -62,6 +63,20 @@ const FileListCommand = cmd({
  },
 })

+const FileTreeCommand = cmd({
+  command: "tree [dir]",
+  builder: (yargs) =>
+    yargs.positional("dir", {
+      type: "string",
+      description: "Directory to tree",
+      default: process.cwd(),
+    }),
+  async handler(args) {
+    const files = await Ripgrep.tree({ cwd: args.dir, limit: 200 })
+    console.log(files)
+  },
+})
+
 export const FileCommand = cmd({
  command: "file",
  builder: (yargs) =>
@@ -70,6 +85,7 @@ export const FileCommand = cmd({
      .command(FileStatusCommand)
      .command(FileListCommand)
      .command(FileSearchCommand)
+      .command(FileTreeCommand)
      .demandCommand(),
  async handler() {},
 })
@@ -312,7 +312,7 @@ function App() {
    {
      title: "Exit the app",
      value: "app.exit",
-      onSelect: exit,
+      onSelect: () => exit(),
      category: "System",
    },
    {
@@ -245,6 +245,11 @@ export function Autocomplete(props: {
          description: "jump to message",
          onSelect: () => command.trigger("session.timeline"),
        },
+        {
+          display: "/thinking",
+          description: "toggle thinking blocks",
+          onSelect: () => command.trigger("session.toggle.thinking"),
+        },
      )
      if (sync.data.config.share !== "disabled") {
        results.push({
@@ -4,7 +4,7 @@ import { onMount } from "solid-js"
 import { createStore, produce } from "solid-js/store"
 import { clone } from "remeda"
 import { createSimpleContext } from "../../context/helper"
-import { appendFile } from "fs/promises"
+import { appendFile, writeFile } from "fs/promises"
 import type { AgentPart, FilePart, TextPart } from "@opencode-ai/sdk"

 export type PromptInfo = {
@@ -24,6 +24,8 @@ export type PromptInfo = {
  )[]
 }

+const MAX_HISTORY_ENTRIES = 50
+
 export const { use: usePromptHistory, provider: PromptHistoryProvider } = createSimpleContext({
  name: "PromptHistory",
  init: () => {
@@ -33,8 +35,23 @@ export const { use: usePromptHistory, provider: PromptHistoryProvider } = create
      const lines = text
        .split("\n")
        .filter(Boolean)
-        .map((line) => JSON.parse(line))
-      setStore("history", lines as PromptInfo[])
+        .map((line) => {
+          try {
+            return JSON.parse(line)
+          } catch {
+            return null
+          }
+        })
+        .filter((line): line is PromptInfo => line !== null)
+        .slice(-MAX_HISTORY_ENTRIES)
+
+      setStore("history", lines)
+
+      // Rewrite file with only valid entries to self-heal corruption
+      if (lines.length > 0) {
+        const content = lines.map((line) => JSON.stringify(line)).join("\n") + "\n"
+        writeFile(historyFile.name!, content).catch(() => {})
+      }
    })

    const [store, setStore] = createStore({
@@ -64,14 +81,26 @@ export const { use: usePromptHistory, provider: PromptHistoryProvider } = create
        return store.history.at(store.index)
      },
      append(item: PromptInfo) {
-        item = clone(item)
-        appendFile(historyFile.name!, JSON.stringify(item) + "\n")
+        const entry = clone(item)
+        let trimmed = false
        setStore(
          produce((draft) => {
-            draft.history.push(item)
+            draft.history.push(entry)
+            if (draft.history.length > MAX_HISTORY_ENTRIES) {
+              draft.history = draft.history.slice(-MAX_HISTORY_ENTRIES)
+              trimmed = true
+            }
            draft.index = 0
          }),
        )
+
+        if (trimmed) {
+          const content = store.history.map((line) => JSON.stringify(line)).join("\n") + "\n"
+          writeFile(historyFile.name!, content).catch(() => {})
+          return
+        }
+
+        appendFile(historyFile.name!, JSON.stringify(entry) + "\n").catch(() => {})
      },
    }
  },
@@ -590,8 +590,7 @@ export function Prompt(props: PromptProps) {
                syncExtmarksWithPromptParts()
              }}
              keyBindings={textareaKeybindings()}
-              // TODO: fix this any
-              onKeyDown={async (e: any) => {
+              onKeyDown={async (e) => {
                if (props.disabled) {
                  e.preventDefault()
                  return
@@ -665,7 +664,11 @@ export function Prompt(props: PromptProps) {
                  return
                }

-                const pastedContent = event.text.trim()
+                // Normalize line endings at the boundary
+                // Windows ConPTY/Terminal often sends CR-only newlines in bracketed paste
+                // Replace CRLF first, then any remaining CR
+                const normalizedText = event.text.replace(/\r\n/g, "\n").replace(/\r/g, "\n")
+                const pastedContent = normalizedText.trim()
                if (!pastedContent) {
                  command.trigger("prompt.paste")
                  return
@@ -1,13 +1,18 @@
 import { useRenderer } from "@opentui/solid"
 import { createSimpleContext } from "./helper"
+import { FormatError } from "@/cli/error"

 export const { use: useExit, provider: ExitProvider } = createSimpleContext({
  name: "Exit",
  init: (input: { onExit?: () => Promise<void> }) => {
    const renderer = useRenderer()
-    return async () => {
+    return async (reason?: any) => {
      renderer.destroy()
      await input.onExit?.()
+      if (reason) {
+        const formatted = FormatError(reason) ?? JSON.stringify(reason)
+        process.stderr.write(formatted + "\n")
+      }
      process.exit(0)
    }
  },
@@ -158,10 +158,10 @@ export const { use: useLocal, provider: LocalProvider } = createSimpleContext({
          }
        }
        const provider = sync.data.provider[0]
-        const model = Object.values(provider.models)[0]
+        const model = sync.data.provider_default[provider.id] ?? Object.values(provider.models)[0].id
        return {
          providerID: provider.id,
-          modelID: model.id,
+          modelID: model,
        }
      })

@@ -11,12 +11,15 @@ import type {
  LspStatus,
  McpStatus,
  FormatterStatus,
+  SessionStatus,
 } from "@opencode-ai/sdk"
 import { createStore, produce, reconcile } from "solid-js/store"
 import { useSDK } from "@tui/context/sdk"
 import { Binary } from "@/util/binary"
 import { createSimpleContext } from "./helper"
 import type { Snapshot } from "@/snapshot"
+import { useExit } from "./exit"
+import { batch, onMount } from "solid-js"

 export const { use: useSync, provider: SyncProvider } = createSimpleContext({
  name: "Sync",
@@ -24,6 +27,7 @@ export const { use: useSync, provider: SyncProvider } = createSimpleContext({
    const [store, setStore] = createStore<{
      status: "loading" | "partial" | "complete"
      provider: Provider[]
+      provider_default: Record<string, string>
      agent: Agent[]
      command: Command[]
      permission: {
@@ -31,6 +35,9 @@ export const { use: useSync, provider: SyncProvider } = createSimpleContext({
      }
      config: Config
      session: Session[]
+      session_status: {
+        [sessionID: string]: SessionStatus
+      }
      session_diff: {
        [sessionID: string]: Snapshot.FileDiff[]
      }
@@ -55,7 +62,9 @@ export const { use: useSync, provider: SyncProvider } = createSimpleContext({
      permission: {},
      command: [],
      provider: [],
+      provider_default: {},
      session: [],
+      session_status: {},
      session_diff: {},
      todo: {},
      message: {},
@@ -138,6 +147,12 @@ export const { use: useSync, provider: SyncProvider } = createSimpleContext({
            }),
          )
          break
+
+        case "session.status": {
+          setStore("session_status", event.properties.sessionID, event.properties.status)
+          break
+        }
+
        case "message.updated": {
          const messages = store.message[event.properties.info.sessionID]
          if (!messages) {
@@ -215,28 +230,42 @@ export const { use: useSync, provider: SyncProvider } = createSimpleContext({
      }
    })

-    // blocking
-    Promise.all([
-      sdk.client.config.providers({ throwOnError: true }).then((x) => setStore("provider", x.data!.providers)),
-      sdk.client.app.agents({ throwOnError: true }).then((x) => setStore("agent", x.data ?? [])),
-      sdk.client.config.get({ throwOnError: true }).then((x) => setStore("config", x.data!)),
-    ]).then(() => {
-      setStore("status", "partial")
-      // non-blocking
+    const exit = useExit()
+
+    onMount(() => {
+      // blocking
      Promise.all([
-        sdk.client.session.list().then((x) =>
-          setStore(
-            "session",
-            (x.data ?? []).toSorted((a, b) => a.id.localeCompare(b.id)),
-          ),
-        ),
-        sdk.client.command.list().then((x) => setStore("command", x.data ?? [])),
-        sdk.client.lsp.status().then((x) => setStore("lsp", x.data!)),
-        sdk.client.mcp.status().then((x) => setStore("mcp", x.data!)),
-        sdk.client.formatter.status().then((x) => setStore("formatter", x.data!)),
-      ]).then(() => {
-        setStore("status", "complete")
-      })
+        sdk.client.config.providers({ throwOnError: true }).then((x) => {
+          batch(() => {
+            setStore("provider", x.data!.providers)
+            setStore("provider_default", x.data!.default)
+          })
+        }),
+        sdk.client.app.agents({ throwOnError: true }).then((x) => setStore("agent", x.data ?? [])),
+        sdk.client.config.get({ throwOnError: true }).then((x) => setStore("config", x.data!)),
+      ])
+        .then(() => {
+          setStore("status", "partial")
+          // non-blocking
+          Promise.all([
+            sdk.client.session.list().then((x) =>
+              setStore(
+                "session",
+                (x.data ?? []).toSorted((a, b) => a.id.localeCompare(b.id)),
+              ),
+            ),
+            sdk.client.command.list().then((x) => setStore("command", x.data ?? [])),
+            sdk.client.lsp.status().then((x) => setStore("lsp", x.data!)),
+            sdk.client.mcp.status().then((x) => setStore("mcp", x.data!)),
+            sdk.client.formatter.status().then((x) => setStore("formatter", x.data!)),
+            sdk.client.session.status().then((x) => setStore("session_status", x.data!)),
+          ]).then(() => {
+            setStore("status", "complete")
+          })
+        })
+        .catch(async (e) => {
+          await exit(e)
+        })
    })

    const result = {
@@ -9,6 +9,7 @@ import catppuccin from "./theme/catppuccin.json" with { type: "json" }
 import cobalt2 from "./theme/cobalt2.json" with { type: "json" }
 import dracula from "./theme/dracula.json" with { type: "json" }
 import everforest from "./theme/everforest.json" with { type: "json" }
+import flexoki from "./theme/flexoki.json" with { type: "json" }
 import github from "./theme/github.json" with { type: "json" }
 import gruvbox from "./theme/gruvbox.json" with { type: "json" }
 import kanagawa from "./theme/kanagawa.json" with { type: "json" }
@@ -105,6 +106,7 @@ export const DEFAULT_THEMES: Record<string, ThemeJson> = {
  cobalt2,
  dracula,
  everforest,
+  flexoki,
  github,
  gruvbox,
  kanagawa,
@@ -128,7 +130,10 @@ function resolveTheme(theme: ThemeJson, mode: "dark" | "light") {
  const defs = theme.defs ?? {}
  function resolveColor(c: ColorValue): RGBA {
    if (c instanceof RGBA) return c
-    if (typeof c === "string") return c.startsWith("#") ? RGBA.fromHex(c) : resolveColor(defs[c])
+    if (typeof c === "string") {
+      if (c === "transparent" || c === "none") return RGBA.fromInts(0, 0, 0, 0)
+      return c.startsWith("#") ? RGBA.fromHex(c) : resolveColor(defs[c])
+    }
    return resolveColor(c[mode])
  }
  return Object.fromEntries(
@@ -864,18 +869,21 @@ function generateSyntax(theme: Theme) {
      scope: ["diff.plus"],
      style: {
        foreground: theme.diffAdded,
+        background: theme.diffAddedBg,
      },
    },
    {
      scope: ["diff.minus"],
      style: {
        foreground: theme.diffRemoved,
+        background: theme.diffRemovedBg,
      },
    },
    {
      scope: ["diff.delta"],
      style: {
        foreground: theme.diffContext,
+        background: theme.diffContextBg,
      },
    },
    {
@@ -0,0 +1,237 @@
+{
+  "$schema": "https://opencode.ai/theme.json",
+  "defs": {
+    "black": "#100F0F",
+    "base950": "#1C1B1A",
+    "base900": "#282726",
+    "base850": "#343331",
+    "base800": "#403E3C",
+    "base700": "#575653",
+    "base600": "#6F6E69",
+    "base500": "#878580",
+    "base300": "#B7B5AC",
+    "base200": "#CECDC3",
+    "base150": "#DAD8CE",
+    "base100": "#E6E4D9",
+    "base50": "#F2F0E5",
+    "paper": "#FFFCF0",
+    "red400": "#D14D41",
+    "red600": "#AF3029",
+    "orange400": "#DA702C",
+    "orange600": "#BC5215",
+    "yellow400": "#D0A215",
+    "yellow600": "#AD8301",
+    "green400": "#879A39",
+    "green600": "#66800B",
+    "cyan400": "#3AA99F",
+    "cyan600": "#24837B",
+    "blue400": "#4385BE",
+    "blue600": "#205EA6",
+    "purple400": "#8B7EC8",
+    "purple600": "#5E409D",
+    "magenta400": "#CE5D97",
+    "magenta600": "#A02F6F"
+  },
+  "theme": {
+    "primary": {
+      "dark": "orange400",
+      "light": "blue600"
+    },
+    "secondary": {
+      "dark": "blue400",
+      "light": "purple600"
+    },
+    "accent": {
+      "dark": "purple400",
+      "light": "orange600"
+    },
+    "error": {
+      "dark": "red400",
+      "light": "red600"
+    },
+    "warning": {
+      "dark": "orange400",
+      "light": "orange600"
+    },
+    "success": {
+      "dark": "green400",
+      "light": "green600"
+    },
+    "info": {
+      "dark": "cyan400",
+      "light": "cyan600"
+    },
+    "text": {
+      "dark": "base200",
+      "light": "black"
+    },
+    "textMuted": {
+      "dark": "base600",
+      "light": "base600"
+    },
+    "background": {
+      "dark": "black",
+      "light": "paper"
+    },
+    "backgroundPanel": {
+      "dark": "base950",
+      "light": "base50"
+    },
+    "backgroundElement": {
+      "dark": "base900",
+      "light": "base100"
+    },
+    "border": {
+      "dark": "base700",
+      "light": "base300"
+    },
+    "borderActive": {
+      "dark": "base600",
+      "light": "base500"
+    },
+    "borderSubtle": {
+      "dark": "base800",
+      "light": "base200"
+    },
+    "diffAdded": {
+      "dark": "green400",
+      "light": "green600"
+    },
+    "diffRemoved": {
+      "dark": "red400",
+      "light": "red600"
+    },
+    "diffContext": {
+      "dark": "base600",
+      "light": "base600"
+    },
+    "diffHunkHeader": {
+      "dark": "blue400",
+      "light": "blue600"
+    },
+    "diffHighlightAdded": {
+      "dark": "green400",
+      "light": "green600"
+    },
+    "diffHighlightRemoved": {
+      "dark": "red400",
+      "light": "red600"
+    },
+    "diffAddedBg": {
+      "dark": "#1A2D1A",
+      "light": "#D5E5D5"
+    },
+    "diffRemovedBg": {
+      "dark": "#2D1A1A",
+      "light": "#F7D8DB"
+    },
+    "diffContextBg": {
+      "dark": "base950",
+      "light": "base50"
+    },
+    "diffLineNumber": {
+      "dark": "base600",
+      "light": "base600"
+    },
+    "diffAddedLineNumberBg": {
+      "dark": "#152515",
+      "light": "#C5D5C5"
+    },
+    "diffRemovedLineNumberBg": {
+      "dark": "#251515",
+      "light": "#E7C8CB"
+    },
+    "markdownText": {
+      "dark": "base200",
+      "light": "black"
+    },
+    "markdownHeading": {
+      "dark": "purple400",
+      "light": "purple600"
+    },
+    "markdownLink": {
+      "dark": "blue400",
+      "light": "blue600"
+    },
+    "markdownLinkText": {
+      "dark": "cyan400",
+      "light": "cyan600"
+    },
+    "markdownCode": {
+      "dark": "cyan400",
+      "light": "cyan600"
+    },
+    "markdownBlockQuote": {
+      "dark": "yellow400",
+      "light": "yellow600"
+    },
+    "markdownEmph": {
+      "dark": "yellow400",
+      "light": "yellow600"
+    },
+    "markdownStrong": {
+      "dark": "orange400",
+      "light": "orange600"
+    },
+    "markdownHorizontalRule": {
+      "dark": "base600",
+      "light": "base600"
+    },
+    "markdownListItem": {
+      "dark": "orange400",
+      "light": "orange600"
+    },
+    "markdownListEnumeration": {
+      "dark": "cyan400",
+      "light": "cyan600"
+    },
+    "markdownImage": {
+      "dark": "magenta400",
+      "light": "magenta600"
+    },
+    "markdownImageText": {
+      "dark": "cyan400",
+      "light": "cyan600"
+    },
+    "markdownCodeBlock": {
+      "dark": "base200",
+      "light": "black"
+    },
+    "syntaxComment": {
+      "dark": "base600",
+      "light": "base600"
+    },
+    "syntaxKeyword": {
+      "dark": "green400",
+      "light": "green600"
+    },
+    "syntaxFunction": {
+      "dark": "orange400",
+      "light": "orange600"
+    },
+    "syntaxVariable": {
+      "dark": "blue400",
+      "light": "blue600"
+    },
+    "syntaxString": {
+      "dark": "cyan400",
+      "light": "cyan600"
+    },
+    "syntaxNumber": {
+      "dark": "purple400",
+      "light": "purple600"
+    },
+    "syntaxType": {
+      "dark": "yellow400",
+      "light": "yellow600"
+    },
+    "syntaxOperator": {
+      "dark": "base300",
+      "light": "base600"
+    },
+    "syntaxPunctuation": {
+      "dark": "base300",
+      "light": "base600"
+    }
+  }
+}
@@ -3,6 +3,7 @@ import { useSync } from "@tui/context/sync"
 import { DialogSelect } from "@tui/ui/dialog-select"
 import { useSDK } from "@tui/context/sdk"
 import { useRoute } from "@tui/context/route"
+import { Clipboard } from "@tui/util/clipboard"
 import type { PromptInfo } from "@tui/component/prompt/history"

 export function DialogMessage(props: {
@@ -54,6 +55,26 @@ export function DialogMessage(props: {
            dialog.clear()
          },
        },
+        {
+          title: "Copy",
+          value: "message.copy",
+          description: "copy message text to clipboard",
+          onSelect: async (dialog) => {
+            const msg = message()
+            if (!msg) return
+
+            const parts = sync.data.part[msg.id]
+            const text = parts.reduce((agg, part) => {
+              if (part.type === "text" && !part.synthetic) {
+                agg += part.text
+              }
+              return agg
+            }, "")
+
+            await Clipboard.copy(text)
+            dialog.clear()
+          },
+        },
        {
          title: "Fork",
          value: "session.fork",
@@ -6,6 +6,8 @@ import {
  For,
  Match,
  on,
+  onCleanup,
+  onMount,
  Show,
  Switch,
  useContext,
@@ -20,7 +22,6 @@ import { useTheme } from "@tui/context/theme"
 import {
  BoxRenderable,
  ScrollBoxRenderable,
-  TextAttributes,
  addDefaultParsers,
  MacOSScrollAccel,
  type ScrollAcceleration,
@@ -65,7 +66,6 @@ import { Editor } from "../../util/editor"
 import { Global } from "@/global"
 import fs from "fs/promises"
 import stripAnsi from "strip-ansi"
-import { LSP } from "@/lsp/index.ts"

 addDefaultParsers(parsers.parsers)

@@ -82,6 +82,7 @@ class CustomSpeedScroll implements ScrollAcceleration {
 const context = createContext<{
  width: number
  conceal: () => boolean
+  showThinking: () => boolean
 }>()

 function use() {
@@ -101,12 +102,18 @@ export function Session() {
  const permissions = createMemo(() => sync.data.permission[route.sessionID] ?? [])

  const pending = createMemo(() => {
-    return messages().findLast((x) => x.role === "assistant" && !x.time?.completed)?.id
+    return messages().findLast((x) => x.role === "assistant" && !x.time.completed)?.id
+  })
+
+  const lastUserMessage = createMemo(() => {
+    const p = pending()
+    return messages().findLast((x) => x.role === "user" && (!p || x.id < p)) as UserMessage
  })

  const dimensions = useTerminalDimensions()
  const [sidebar, setSidebar] = createSignal<"show" | "hide" | "auto">(kv.get("sidebar", "auto"))
  const [conceal, setConceal] = createSignal(true)
+  const [showThinking, setShowThinking] = createSignal(true)

  const wide = createMemo(() => dimensions().width > 120)
  const sidebarVisible = createMemo(() => sidebar() === "show" || (sidebar() === "auto" && wide()))
@@ -380,6 +387,15 @@ export function Session() {
        dialog.clear()
      },
    },
+    {
+      title: "Toggle thinking blocks",
+      value: "session.toggle.thinking",
+      category: "Session",
+      onSelect: (dialog) => {
+        setShowThinking((prev) => !prev)
+        dialog.clear()
+      },
+    },
    {
      title: "Page up",
      value: "session.page.up",
@@ -669,6 +685,7 @@ export function Session() {
          return contentWidth()
        },
        conceal,
+        showThinking,
      }}
    >
      <box flexDirection="row" paddingBottom={1} paddingTop={1} paddingLeft={2} paddingRight={2} gap={2}>
@@ -801,7 +818,7 @@ export function Session() {
                    </Match>
                    <Match when={message.role === "assistant"}>
                      <AssistantMessage
-                        last={index() === messages().length - 1}
+                        last={pending() === message.id}
                        message={message as AssistantMessage}
                        parts={sync.data.part[message.id] ?? []}
                      />
@@ -856,64 +873,84 @@ function UserMessage(props: {
  const queued = createMemo(() => props.pending && props.message.id > props.pending)
  const color = createMemo(() => (queued() ? theme.accent : theme.secondary))

+  const compaction = createMemo(() => props.parts.find((x) => x.type === "compaction"))
+
  return (
-    <Show when={text()}>
-      <box
-        id={props.message.id}
-        onMouseOver={() => {
-          setHover(true)
-        }}
-        onMouseOut={() => {
-          setHover(false)
-        }}
-        onMouseUp={props.onMouseUp}
-        border={["left"]}
-        paddingTop={1}
-        paddingBottom={1}
-        paddingLeft={2}
-        marginTop={props.index === 0 ? 0 : 1}
-        backgroundColor={hover() ? theme.backgroundElement : theme.backgroundPanel}
-        customBorderChars={SplitBorder.customBorderChars}
-        borderColor={color()}
-        flexShrink={0}
-      >
-        <text fg={theme.text}>{text()?.text}</text>
-        <Show when={files().length}>
-          <box flexDirection="row" paddingBottom={1} paddingTop={1} gap={1} flexWrap="wrap">
-            <For each={files()}>
-              {(file) => {
-                const bg = createMemo(() => {
-                  if (file.mime.startsWith("image/")) return theme.accent
-                  if (file.mime === "application/pdf") return theme.primary
-                  return theme.secondary
-                })
-                return (
-                  <text fg={theme.text}>
-                    <span style={{ bg: bg(), fg: theme.background }}> {MIME_BADGE[file.mime] ?? file.mime} </span>
-                    <span style={{ bg: theme.backgroundElement, fg: theme.textMuted }}> {file.filename} </span>
-                  </text>
-                )
-              }}
-            </For>
-          </box>
-        </Show>
-        <text fg={theme.text}>
-          {sync.data.config.username ?? "You"}{" "}
-          <Show
-            when={queued()}
-            fallback={<span style={{ fg: theme.textMuted }}>({Locale.time(props.message.time.created)})</span>}
-          >
-            <span style={{ bg: theme.accent, fg: theme.backgroundPanel, bold: true }}> QUEUED </span>
+    <>
+      <Show when={text()}>
+        <box
+          id={props.message.id}
+          onMouseOver={() => {
+            setHover(true)
+          }}
+          onMouseOut={() => {
+            setHover(false)
+          }}
+          onMouseUp={props.onMouseUp}
+          border={["left"]}
+          paddingTop={1}
+          paddingBottom={1}
+          paddingLeft={2}
+          marginTop={props.index === 0 ? 0 : 1}
+          backgroundColor={hover() ? theme.backgroundElement : theme.backgroundPanel}
+          customBorderChars={SplitBorder.customBorderChars}
+          borderColor={color()}
+          flexShrink={0}
+        >
+          <text fg={theme.text}>{text()?.text}</text>
+          <Show when={files().length}>
+            <box flexDirection="row" paddingBottom={1} paddingTop={1} gap={1} flexWrap="wrap">
+              <For each={files()}>
+                {(file) => {
+                  const bg = createMemo(() => {
+                    if (file.mime.startsWith("image/")) return theme.accent
+                    if (file.mime === "application/pdf") return theme.primary
+                    return theme.secondary
+                  })
+                  return (
+                    <text fg={theme.text}>
+                      <span style={{ bg: bg(), fg: theme.background }}> {MIME_BADGE[file.mime] ?? file.mime} </span>
+                      <span style={{ bg: theme.backgroundElement, fg: theme.textMuted }}> {file.filename} </span>
+                    </text>
+                  )
+                }}
+              </For>
+            </box>
          </Show>
-        </text>
-      </box>
-    </Show>
+          <text fg={theme.text}>
+            {sync.data.config.username ?? "You"}{" "}
+            <Show
+              when={queued()}
+              fallback={<span style={{ fg: theme.textMuted }}>({Locale.time(props.message.time.created)})</span>}
+            >
+              <span style={{ bg: theme.accent, fg: theme.backgroundPanel, bold: true }}> QUEUED </span>
+            </Show>
+          </text>
+        </box>
+      </Show>
+      <Show when={compaction()}>
+        <box
+          marginTop={1}
+          border={["top"]}
+          title=" Compaction "
+          titleAlignment="center"
+          borderColor={theme.borderActive}
+        />
+      </Show>
+    </>
  )
 }

 function AssistantMessage(props: { message: AssistantMessage; parts: Part[]; last: boolean }) {
  const local = useLocal()
  const { theme } = useTheme()
+  const sync = useSync()
+  const status = createMemo(
+    () =>
+      sync.data.session_status[props.message.sessionID] ?? {
+        type: "idle",
+      },
+  )
  return (
    <>
      <For each={props.parts}>
@@ -945,23 +982,36 @@ function AssistantMessage(props: { message: AssistantMessage; parts: Part[]; las
          <text fg={theme.textMuted}>{props.message.error?.data.message}</text>
        </box>
      </Show>
-      <Show
-        when={
-          !props.message.time.completed ||
-          (props.last && props.parts.some((item) => item.type === "step-finish" && item.reason === "tool-calls"))
-        }
-      >
-        <box
-          paddingLeft={2}
-          marginTop={1}
-          flexDirection="row"
-          gap={1}
-          border={["left"]}
-          customBorderChars={SplitBorder.customBorderChars}
-          borderColor={theme.backgroundElement}
-        >
+      <Show when={props.last && status().type !== "idle"}>
+        <box paddingLeft={3} flexDirection="row" gap={1} marginTop={1}>
          <text fg={local.agent.color(props.message.mode)}>{Locale.titlecase(props.message.mode)}</text>
-          <Shimmer text={`${props.message.modelID}`} color={theme.text} />
+          <Shimmer text={props.message.modelID} color={theme.text} />
+          {(() => {
+            const retry = createMemo(() => {
+              const s = status()
+              if (s.type !== "retry") return
+              return s
+            })
+            const [seconds, setSeconds] = createSignal(0)
+            onMount(() => {
+              const timer = setInterval(() => {
+                const next = retry()?.next
+                if (next) setSeconds(Math.round((next - Date.now()) / 1000))
+              }, 1000)
+
+              onCleanup(() => {
+                clearInterval(timer)
+              })
+            })
+            return (
+              <Show when={retry()}>
+                <text fg={theme.error}>
+                  {retry()!.message} [attempt #{retry()!.attempt}
+                  {seconds() > 0 ? `, retrying in ${seconds()}s` : ""}]
+                </text>
+              </Show>
+            )
+          })()}
        </box>
      </Show>
      <Show
@@ -992,7 +1042,7 @@ function ReasoningPart(props: { last: boolean; part: ReasoningPart; message: Ass
  const ctx = use()
  const content = createMemo(() => props.part.text.trim())
  return (
-    <Show when={content()}>
+    <Show when={content() && ctx.showThinking()}>
      <box
        id={"text-" + props.part.id}
        paddingLeft={2}
@@ -1225,9 +1275,7 @@ ToolRegistry.register<typeof WriteTool>({
  container: "block",
  render(props) {
    const { theme, syntax } = useTheme()
-    const lines = createMemo(() => {
-      return props.input.content?.split("\n") ?? []
-    })
+    const lines = createMemo(() => props.input.content?.split("\n") ?? [], [] as string[])
    const code = createMemo(() => {
      if (!props.input.content) return ""
      const text = props.input.content
@@ -60,13 +60,19 @@ export function Sidebar(props: { sessionID: string }) {
          </box>
          <Show when={Object.keys(sync.data.mcp).length > 0}>
            <box>
-              <box flexDirection="row" gap={1} onMouseDown={() => setMcpExpanded(!mcpExpanded())}>
-                <text fg={theme.text}>{mcpExpanded() ? "▼" : "▶"}</text>
+              <box
+                flexDirection="row"
+                gap={1}
+                onMouseDown={() => Object.keys(sync.data.mcp).length > 2 && setMcpExpanded(!mcpExpanded())}
+              >
+                <Show when={Object.keys(sync.data.mcp).length > 2}>
+                  <text fg={theme.text}>{mcpExpanded() ? "▼" : "▶"}</text>
+                </Show>
                <text fg={theme.text}>
                  <b>MCP</b>
                </text>
              </box>
-              <Show when={mcpExpanded()}>
+              <Show when={Object.keys(sync.data.mcp).length <= 2 || mcpExpanded()}>
                <For each={Object.entries(sync.data.mcp)}>
                  {([key, item]) => (
                    <box flexDirection="row" gap={1}>
@@ -100,13 +106,19 @@ export function Sidebar(props: { sessionID: string }) {
          </Show>
          <Show when={sync.data.lsp.length > 0}>
            <box>
-              <box flexDirection="row" gap={1} onMouseDown={() => setLspExpanded(!lspExpanded())}>
-                <text fg={theme.text}>{lspExpanded() ? "▼" : "▶"}</text>
+              <box
+                flexDirection="row"
+                gap={1}
+                onMouseDown={() => sync.data.lsp.length > 2 && setLspExpanded(!lspExpanded())}
+              >
+                <Show when={sync.data.lsp.length > 2}>
+                  <text fg={theme.text}>{lspExpanded() ? "▼" : "▶"}</text>
+                </Show>
                <text fg={theme.text}>
                  <b>LSP</b>
                </text>
              </box>
-              <Show when={lspExpanded()}>
+              <Show when={sync.data.lsp.length <= 2 || lspExpanded()}>
                <For each={sync.data.lsp}>
                  {(item) => (
                    <box flexDirection="row" gap={1}>
@@ -132,13 +144,19 @@ export function Sidebar(props: { sessionID: string }) {
          </Show>
          <Show when={todo().length > 0}>
            <box>
-              <box flexDirection="row" gap={1} onMouseDown={() => setTodoExpanded(!todoExpanded())}>
-                <text fg={theme.text}>{todoExpanded() ? "▼" : "▶"}</text>
+              <box
+                flexDirection="row"
+                gap={1}
+                onMouseDown={() => todo().length > 2 && setTodoExpanded(!todoExpanded())}
+              >
+                <Show when={todo().length > 2}>
+                  <text fg={theme.text}>{todoExpanded() ? "▼" : "▶"}</text>
+                </Show>
                <text fg={theme.text}>
                  <b>Todo</b>
                </text>
              </box>
-              <Show when={todoExpanded()}>
+              <Show when={todo().length <= 2 || todoExpanded()}>
                <For each={todo()}>
                  {(todo) => (
                    <text style={{ fg: todo.status === "in_progress" ? theme.success : theme.textMuted }}>
@@ -151,13 +169,19 @@ export function Sidebar(props: { sessionID: string }) {
          </Show>
          <Show when={diff().length > 0}>
            <box>
-              <box flexDirection="row" gap={1} onMouseDown={() => setDiffExpanded(!diffExpanded())}>
-                <text fg={theme.text}>{diffExpanded() ? "▼" : "▶"}</text>
+              <box
+                flexDirection="row"
+                gap={1}
+                onMouseDown={() => diff().length > 2 && setDiffExpanded(!diffExpanded())}
+              >
+                <Show when={diff().length > 2}>
+                  <text fg={theme.text}>{diffExpanded() ? "▼" : "▶"}</text>
+                </Show>
                <text fg={theme.text}>
                  <b>Modified Files</b>
                </text>
              </box>
-              <Show when={diffExpanded()}>
+              <Show when={diff().length <= 2 || diffExpanded()}>
                <For each={diff() || []}>
                  {(item) => {
                    const file = createMemo(() => {
@@ -2,10 +2,12 @@ import { TextAttributes } from "@opentui/core"
 import { useTheme } from "@tui/context/theme"
 import { useDialog } from "./dialog"
 import { useKeyboard } from "@opentui/solid"
+import { useKeybind } from "@tui/context/keybind"

 export function DialogHelp() {
  const dialog = useDialog()
  const { theme } = useTheme()
+  const keybind = useKeybind()

  useKeyboard((evt) => {
    if (evt.name === "return" || evt.name === "escape") {
@@ -20,7 +22,9 @@ export function DialogHelp() {
        <text fg={theme.textMuted}>esc/enter</text>
      </box>
      <box paddingBottom={1}>
-        <text fg={theme.textMuted}>Press Ctrl+P to see all available actions and commands in any context.</text>
+        <text fg={theme.textMuted}>
+          Press {keybind.print("command_list")} to see all available actions and commands in any context.
+        </text>
      </box>
      <box flexDirection="row" justifyContent="flex-end" paddingBottom={1}>
        <box paddingLeft={3} paddingRight={3} backgroundColor={theme.primary} onMouseUp={() => dialog.clear()}>
@@ -43,6 +43,7 @@ export const rpc = {
    }
  },
  async shutdown() {
+    Log.Default.info("worker shutting down")
    await Instance.disposeAll()
    await server.stop(true)
  },
@@ -75,7 +75,7 @@ export namespace Config {
    for (const dir of directories) {
      await assertValid(dir)

-      if (dir.endsWith(".opencode")) {
+      if (dir.endsWith(".opencode") || dir === Flag.OPENCODE_CONFIG_DIR) {
        for (const file of ["opencode.jsonc", "opencode.json"]) {
          log.debug(`loading config from ${path.join(dir, file)}`)
          result = mergeDeep(result, await loadFile(path.join(dir, file)))
@@ -337,7 +337,7 @@ export namespace Config {
  export const Mcp = z.discriminatedUnion("type", [McpLocal, McpRemote])
  export type Mcp = z.infer<typeof Mcp>

-  export const Permission = z.union([z.literal("ask"), z.literal("allow"), z.literal("deny")])
+  export const Permission = z.enum(["ask", "allow", "deny"])
  export type Permission = z.infer<typeof Permission>

  export const Command = z.object({
@@ -358,7 +358,7 @@ export namespace Config {
      tools: z.record(z.string(), z.boolean()).optional(),
      disable: z.boolean().optional(),
      description: z.string().optional().describe("Description of when to use the agent"),
-      mode: z.union([z.literal("subagent"), z.literal("primary"), z.literal("all")]).optional(),
+      mode: z.enum(["subagent", "primary", "all"]).optional(),
      color: z
        .string()
        .regex(/^#[0-9a-fA-F]{6}$/, "Invalid hex color format")
@@ -437,7 +437,7 @@ export namespace Config {
    })

  export const TUI = z.object({
-    scroll_speed: z.number().min(1).optional().default(1).describe("TUI scroll speed"),
+    scroll_speed: z.number().min(0.001).optional().default(1).describe("TUI scroll speed"),
    scroll_acceleration: z
      .object({
        enabled: z.boolean().describe("Enable scroll acceleration"),
@@ -622,6 +622,7 @@ export namespace Config {
            .optional(),
          chatMaxRetries: z.number().optional().describe("Number of retries for chat completions on failure"),
          disable_paste_summary: z.boolean().optional(),
+          batch_tool: z.boolean().optional().describe("Enable the batch tool"),
        })
        .optional(),
    })
@@ -8,8 +8,10 @@ import { lazy } from "../util/lazy"
 import { $ } from "bun"

 import { ZipReader, BlobReader, BlobWriter } from "@zip.js/zip.js"
+import { Log } from "@/util/log"

 export namespace Ripgrep {
+  const log = Log.create({ service: "ripgrep" })
  const Stats = z.object({
    elapsed: z.object({
      secs: z.number(),
@@ -254,6 +256,7 @@ export namespace Ripgrep {
  }

  export async function tree(input: { cwd: string; limit?: number }) {
+    log.info("tree", input)
    const files = await Array.fromAsync(Ripgrep.files({ cwd: input.cwd }))
    interface Node {
      path: string[]
@@ -53,10 +53,14 @@ export const Instance = {
    await State.dispose(Instance.directory)
  },
  async disposeAll() {
+    Log.Default.info("disposing all instances")
    for (const [_key, value] of cache) {
-      await context.provide(await value, async () => {
-        await Instance.dispose()
-      })
+      const awaited = await value.catch(() => {})
+      if (awaited) {
+        await context.provide(await value, async () => {
+          await Instance.dispose()
+        })
+      }
    }
    cache.clear()
  },
@@ -23,6 +23,14 @@ export namespace ModelsDev {
        output: z.number(),
        cache_read: z.number().optional(),
        cache_write: z.number().optional(),
+        context_over_200k: z
+          .object({
+            input: z.number(),
+            output: z.number(),
+            cache_read: z.number().optional(),
+            cache_write: z.number().optional(),
+          })
+          .optional(),
      }),
      limit: z.object({
        context: z.number(),
@@ -53,7 +53,7 @@ export namespace Provider {

      return {
        autoload: Object.keys(input.models).length > 0,
-        options: {},
+        options: hasKey ? {} : { apiKey: "public" },
      }
    },
    openai: async () => {
@@ -78,6 +78,22 @@ export namespace Provider {
        options: {},
      }
    },
+    "azure-cognitive-services": async () => {
+      const resourceName = process.env["AZURE_COGNITIVE_SERVICES_RESOURCE_NAME"]
+      return {
+        autoload: false,
+        async getModel(sdk: any, modelID: string, options?: Record<string, any>) {
+          if (options?.["useCompletionUrls"]) {
+            return sdk.chat(modelID)
+          } else {
+            return sdk.responses(modelID)
+          }
+        },
+        options: {
+          baseURL: resourceName ? `https://${resourceName}.cognitiveservices.azure.com/openai` : undefined,
+        },
+      }
+    },
    "amazon-bedrock": async () => {
      if (!process.env["AWS_PROFILE"] && !process.env["AWS_ACCESS_KEY_ID"] && !process.env["AWS_BEARER_TOKEN_BEDROCK"])
        return { autoload: false }
@@ -209,6 +225,17 @@ export namespace Provider {
        },
      }
    },
+    zenmux: async () => {
+      return {
+        autoload: false,
+        options: {
+          headers: {
+            "HTTP-Referer": "https://opencode.ai/",
+            "X-Title": "opencode",
+          },
+        },
+      }
+    },
  }

  const state = Instance.state(async () => {
@@ -470,7 +497,15 @@ export namespace Provider {
      const key = Bun.hash.xxHash32(JSON.stringify({ pkg, options }))
      const existing = s.sdk.get(key)
      if (existing) return existing
-      const installedPath = await BunProc.install(pkg, "latest")
+
+      let installedPath: string
+      if (!pkg.startsWith("file://")) {
+        installedPath = await BunProc.install(pkg, "latest")
+      } else {
+        log.info("loading local provider", { pkg })
+        installedPath = pkg
+      }
+
      // The `google-vertex-anthropic` provider points to the `@ai-sdk/google-vertex` package.
      // Ref: https://github.com/sst/models.dev/blob/0a87de42ab177bebad0620a889e2eb2b4a5dd4ab/providers/google-vertex-anthropic/provider.toml
      // However, the actual export is at the subpath `@ai-sdk/google-vertex/anthropic`.
@@ -582,6 +617,9 @@ export namespace Provider {
    if (providerID === "github-copilot") {
      priority = priority.filter((m) => m !== "claude-haiku-4.5")
    }
+    if (providerID === "opencode" || providerID === "local") {
+      priority = ["gpt-5-nano"]
+    }
    for (const item of priority) {
      for (const model of Object.keys(provider.info.models)) {
        if (model.includes(item)) return getModel(providerID, model)
@@ -589,7 +627,7 @@ export namespace Provider {
    }
  }

-  const priority = ["gemini-2.5-pro-preview", "gpt-5", "claude-sonnet-4"]
+  const priority = ["gemini-2.5-pro-preview", "gpt-5", "claude-sonnet-4", "big-pickle"]
  export function sort(models: ModelsDev.Model[]) {
    return sortBy(
      models,
@@ -128,7 +128,12 @@ export namespace ProviderTransform {
    return undefined
  }

-  export function options(providerID: string, modelID: string, sessionID: string): Record<string, any> | undefined {
+  export function options(
+    providerID: string,
+    modelID: string,
+    npm: string,
+    sessionID: string,
+  ): Record<string, any> | undefined {
    const result: Record<string, any> = {}

    if (providerID === "openai") {
@@ -144,6 +149,10 @@ export namespace ProviderTransform {
        result["reasoningEffort"] = "medium"
      }

+      if (modelID.endsWith("gpt-5.1") && providerID !== "azure") {
+        result["textVerbosity"] = "low"
+      }
+
      if (providerID === "opencode") {
        result["promptCacheKey"] = sessionID
        result["include"] = ["reasoning.encrypted_content"]
@@ -168,6 +177,10 @@ export namespace ProviderTransform {
        return {
          ["anthropic" as string]: options,
        }
+      case "@ai-sdk/gateway":
+        return {
+          ["gateway" as string]: options,
+        }
      default:
        return {
          [providerID]: options,
@@ -176,7 +189,7 @@ export namespace ProviderTransform {
  }

  export function maxOutputTokens(
-    providerID: string,
+    npm: string,
    options: Record<string, any>,
    modelLimit: number,
    globalLimit: number,
@@ -184,7 +197,7 @@ export namespace ProviderTransform {
    const modelCap = modelLimit || globalLimit
    const standardLimit = Math.min(modelCap, globalLimit)

-    if (providerID === "anthropic") {
+    if (npm === "@ai-sdk/anthropic") {
      const thinking = options?.["thinking"]
      const budgetTokens = typeof thinking?.["budgetTokens"] === "number" ? thinking["budgetTokens"] : 0
      const enabled = thinking?.["type"] === "enabled"
@@ -27,7 +27,6 @@ import { Global } from "../global"
 import { ProjectRoute } from "./project"
 import { ToolRegistry } from "../tool/registry"
 import { zodToJsonSchema } from "zod-to-json-schema"
-import { SessionLock } from "../session/lock"
 import { SessionPrompt } from "../session/prompt"
 import { SessionCompaction } from "../session/compaction"
 import { SessionRevert } from "../session/revert"
@@ -40,6 +39,8 @@ import type { ContentfulStatusCode } from "hono/utils/http-status"
 import { TuiEvent } from "@/cli/cmd/tui/event"
 import { Snapshot } from "@/snapshot"
 import { SessionSummary } from "@/session/summary"
+import { GlobalBus } from "@/bus/global"
+import { SessionStatus } from "@/session/status"

 const ERRORS = {
  400: {
@@ -49,7 +50,7 @@ const ERRORS = {
        schema: resolver(
          z
            .object({
-              data: z.any().nullable(),
+              data: z.any(),
              errors: z.array(z.record(z.string(), z.any())),
              success: z.literal(false),
            })
@@ -117,6 +118,56 @@ export namespace Server {
          timer.stop()
        }
      })
+      .get(
+        "/global/event",
+        describeRoute({
+          description: "Get events",
+          operationId: "global.event",
+          responses: {
+            200: {
+              description: "Event stream",
+              content: {
+                "text/event-stream": {
+                  schema: resolver(
+                    z
+                      .object({
+                        directory: z.string(),
+                        payload: Bus.payloads(),
+                      })
+                      .meta({
+                        ref: "GlobalEvent",
+                      }),
+                  ),
+                },
+              },
+            },
+          },
+        }),
+        async (c) => {
+          log.info("global event connected")
+          return streamSSE(c, async (stream) => {
+            stream.writeSSE({
+              data: JSON.stringify({
+                type: "server.connected",
+                properties: {},
+              }),
+            })
+            async function handler(event: any) {
+              await stream.writeSSE({
+                data: JSON.stringify(event),
+              })
+            }
+            GlobalBus.on("event", handler)
+            await new Promise<void>((resolve) => {
+              stream.onAbort(() => {
+                GlobalBus.off("event", handler)
+                resolve()
+                log.info("global event disconnected")
+              })
+            })
+          })
+        },
+      )
      .use(async (c, next) => {
        const directory = c.req.query("directory") ?? process.cwd()
        return Instance.provide({
@@ -316,6 +367,28 @@ export namespace Server {
          return c.json(sessions)
        },
      )
+      .get(
+        "/session/status",
+        describeRoute({
+          description: "Get session status",
+          operationId: "session.status",
+          responses: {
+            200: {
+              description: "Get session status",
+              content: {
+                "application/json": {
+                  schema: resolver(z.record(z.string(), SessionStatus.Info)),
+                },
+              },
+            },
+            ...errors(400),
+          },
+        }),
+        async (c) => {
+          const result = SessionStatus.list()
+          return c.json(result)
+        },
+      )
      .get(
        "/session/:id",
        describeRoute({
@@ -586,7 +659,8 @@ export namespace Server {
          }),
        ),
        async (c) => {
-          return c.json(SessionLock.abort(c.req.valid("param").id))
+          SessionPrompt.cancel(c.req.valid("param").id)
+          return c.json(true)
        },
      )
      .post(
@@ -720,7 +794,14 @@ export namespace Server {
        async (c) => {
          const id = c.req.valid("param").id
          const body = c.req.valid("json")
-          await SessionCompaction.run({ ...body, sessionID: id })
+          await SessionCompaction.create({
+            sessionID: id,
+            model: {
+              providerID: body.providerID,
+              modelID: body.modelID,
+            },
+          })
+          await SessionPrompt.loop(id)
          return c.json(true)
        },
      )
@@ -1137,7 +1218,7 @@ export namespace Server {
          "query",
          z.object({
            query: z.string(),
-            dirs: z.union([z.literal("true"), z.literal("false")]).optional(),
+            dirs: z.enum(["true", "false"]).optional(),
          }),
        ),
        async (c) => {
@@ -1721,11 +1802,7 @@ export namespace Server {
              description: "Event stream",
              content: {
                "text/event-stream": {
-                  schema: resolver(
-                    Bus.payloads().meta({
-                      ref: "Event",
-                    }),
-                  ),
+                  schema: resolver(Bus.payloads()),
                },
              },
            },
@@ -1,9 +1,8 @@
-import { streamText, type ModelMessage, type StreamTextResult, type Tool as AITool } from "ai"
+import { streamText, type ModelMessage } from "ai"
 import { Session } from "."
 import { Identifier } from "../id/id"
 import { Instance } from "../project/instance"
 import { Provider } from "../provider/provider"
-import { defer } from "../util/defer"
 import { MessageV2 } from "./message-v2"
 import { SystemPrompt } from "./system"
 import { Bus } from "../bus"
@@ -13,10 +12,9 @@ import { SessionPrompt } from "./prompt"
 import { Flag } from "../flag/flag"
 import { Token } from "../util/token"
 import { Log } from "../util/log"
-import { SessionLock } from "./lock"
 import { ProviderTransform } from "@/provider/transform"
-import { SessionRetry } from "./retry"
-import { Config } from "@/config/config"
+import { SessionProcessor } from "./processor"
+import { fn } from "@/util/fn"

 export namespace SessionCompaction {
  const log = Log.create({ service: "session.compaction" })
@@ -42,7 +40,6 @@ export namespace SessionCompaction {

  export const PRUNE_MINIMUM = 20_000
  export const PRUNE_PROTECT = 40_000
-  const MAX_RETRIES = 10

  // goes backwards through parts until there are 40_000 tokens worth of tool
  // calls. then erases output of previous tool calls. idea is to throw away old
@@ -87,38 +84,29 @@ export namespace SessionCompaction {
    }
  }

-  export async function run(input: { sessionID: string; providerID: string; modelID: string; signal?: AbortSignal }) {
-    if (!input.signal) SessionLock.assertUnlocked(input.sessionID)
-    await using lock = input.signal === undefined ? SessionLock.acquire({ sessionID: input.sessionID }) : undefined
-    const signal = input.signal ?? lock!.signal
-
-    await Session.update(input.sessionID, (draft) => {
-      draft.time.compacting = Date.now()
-    })
-    await using _ = defer(async () => {
-      await Session.update(input.sessionID, (draft) => {
-        draft.time.compacting = undefined
-      })
-    })
-    const toSummarize = await MessageV2.filterCompacted(MessageV2.stream(input.sessionID))
-    const model = await Provider.getModel(input.providerID, input.modelID)
-    const system = [
-      ...SystemPrompt.summarize(model.providerID),
-      ...(await SystemPrompt.environment()),
-      ...(await SystemPrompt.custom()),
-    ]
-
+  export async function process(input: {
+    parentID: string
+    messages: MessageV2.WithParts[]
+    sessionID: string
+    model: {
+      providerID: string
+      modelID: string
+    }
+    abort: AbortSignal
+  }) {
+    const model = await Provider.getModel(input.model.providerID, input.model.modelID)
+    const system = [...SystemPrompt.summarize(model.providerID)]
    const msg = (await Session.updateMessage({
      id: Identifier.ascending("message"),
      role: "assistant",
-      parentID: toSummarize.findLast((m) => m.info.role === "user")?.info.id!,
+      parentID: input.parentID,
      sessionID: input.sessionID,
      mode: "build",
+      summary: true,
      path: {
        cwd: Instance.directory,
        root: Instance.worktree,
      },
-      summary: true,
      cost: 0,
      tokens: {
        output: 0,
@@ -126,37 +114,27 @@ export namespace SessionCompaction {
        reasoning: 0,
        cache: { read: 0, write: 0 },
      },
-      modelID: input.modelID,
+      modelID: input.model.modelID,
      providerID: model.providerID,
      time: {
        created: Date.now(),
      },
    })) as MessageV2.Assistant
-
-    const part = (await Session.updatePart({
-      type: "text",
+    const processor = SessionProcessor.create({
+      assistantMessage: msg,
      sessionID: input.sessionID,
-      messageID: msg.id,
-      id: Identifier.ascending("part"),
-      text: "",
-      time: {
-        start: Date.now(),
-      },
-    })) as MessageV2.TextPart
-
-    const doStream = () =>
+      providerID: input.model.providerID,
+      model: model.info,
+      abort: input.abort,
+    })
+    const result = await processor.process(() =>
      streamText({
        // set to 0, we handle loop
        maxRetries: 0,
        model: model.language,
        providerOptions: ProviderTransform.providerOptions(model.npm, model.providerID, model.info.options),
        headers: model.info.headers,
-        abortSignal: signal,
-        onError(error) {
-          log.error("stream error", {
-            error,
-          })
-        },
+        abortSignal: input.abort,
        tools: model.info.tool_call ? {} : undefined,
        messages: [
          ...system.map(
@@ -165,7 +143,7 @@ export namespace SessionCompaction {
              content: x,
            }),
          ),
-          ...MessageV2.toModelMessage(toSummarize),
+          ...MessageV2.toModelMessage(input.messages),
          {
            role: "user",
            content: [
@@ -176,168 +154,60 @@ export namespace SessionCompaction {
            ],
          },
        ],
-      })
-
-    // TODO: reduce duplication between compaction.ts & prompt.ts
-    const process = async (
-      stream: StreamTextResult<Record<string, AITool>, never>,
-      retries: { count: number; max: number },
-    ) => {
-      let shouldRetry = false
-      try {
-        for await (const value of stream.fullStream) {
-          signal.throwIfAborted()
-          switch (value.type) {
-            case "text-delta":
-              part.text += value.text
-              if (value.providerMetadata) part.metadata = value.providerMetadata
-              if (part.text)
-                await Session.updatePart({
-                  part,
-                  delta: value.text,
-                })
-              continue
-            case "text-end": {
-              part.text = part.text.trimEnd()
-              part.time = {
-                start: Date.now(),
-                end: Date.now(),
-              }
-              if (value.providerMetadata) part.metadata = value.providerMetadata
-              await Session.updatePart(part)
-              continue
-            }
-            case "finish-step": {
-              const usage = Session.getUsage({
-                model: model.info,
-                usage: value.usage,
-                metadata: value.providerMetadata,
-              })
-              msg.cost += usage.cost
-              msg.tokens = usage.tokens
-              await Session.updateMessage(msg)
-              continue
-            }
-            case "error":
-              throw value.error
-            default:
-              continue
-          }
-        }
-      } catch (e) {
-        log.error("compaction error", {
-          error: e,
-        })
-        const error = MessageV2.fromError(e, { providerID: input.providerID })
-        if (retries.count < retries.max && MessageV2.APIError.isInstance(error) && error.data.isRetryable) {
-          shouldRetry = true
-          await Session.updatePart({
-            id: Identifier.ascending("part"),
-            messageID: msg.id,
-            sessionID: msg.sessionID,
-            type: "retry",
-            attempt: retries.count + 1,
-            time: {
-              created: Date.now(),
-            },
-            error,
-          })
-        } else {
-          msg.error = error
-          Bus.publish(Session.Event.Error, {
-            sessionID: msg.sessionID,
-            error: msg.error,
-          })
-        }
-      }
-
-      const parts = await MessageV2.parts(msg.id)
-      return {
-        info: msg,
-        parts,
-        shouldRetry,
-      }
-    }
-
-    let stream = doStream()
-    const cfg = await Config.get()
-    const maxRetries = cfg.experimental?.chatMaxRetries ?? MAX_RETRIES
-    let result = await process(stream, {
-      count: 0,
-      max: maxRetries,
-    })
-    if (result.shouldRetry) {
-      const start = Date.now()
-      for (let retry = 1; retry < maxRetries; retry++) {
-        const lastRetryPart = result.parts.findLast((p): p is MessageV2.RetryPart => p.type === "retry")
-
-        if (lastRetryPart) {
-          const delayMs = SessionRetry.getBoundedDelay({
-            error: lastRetryPart.error,
-            attempt: retry,
-            startTime: start,
-          })
-          if (!delayMs) {
-            break
-          }
-
-          log.info("retrying with backoff", {
-            attempt: retry,
-            delayMs,
-            elapsed: Date.now() - start,
-          })
-
-          const stop = await SessionRetry.sleep(delayMs, signal)
-            .then(() => false)
-            .catch((error) => {
-              if (error instanceof DOMException && error.name === "AbortError") {
-                const err = new MessageV2.AbortedError(
-                  { message: error.message },
-                  {
-                    cause: error,
-                  },
-                ).toObject()
-                result.info.error = err
-                Bus.publish(Session.Event.Error, {
-                  sessionID: result.info.sessionID,
-                  error: result.info.error,
-                })
-                return true
-              }
-              throw error
-            })
-
-          if (stop) break
-        }
-
-        stream = doStream()
-        result = await process(stream, {
-          count: retry,
-          max: maxRetries,
-        })
-        if (!result.shouldRetry) {
-          break
-        }
-      }
-    }
-
-    msg.time.completed = Date.now()
-
-    if (
-      !msg.error ||
-      (MessageV2.AbortedError.isInstance(msg.error) &&
-        result.parts.some((part): part is MessageV2.TextPart => part.type === "text" && part.text.length > 0))
-    ) {
-      msg.summary = true
-      Bus.publish(Event.Compacted, {
+      }),
+    )
+    if (result === "continue") {
+      const continueMsg = await Session.updateMessage({
+        id: Identifier.ascending("message"),
+        role: "user",
        sessionID: input.sessionID,
+        time: {
+          created: Date.now(),
+        },
+        agent: "build",
+        model: input.model,
+      })
+      await Session.updatePart({
+        id: Identifier.ascending("part"),
+        messageID: continueMsg.id,
+        sessionID: input.sessionID,
+        type: "text",
+        synthetic: true,
+        text: "Continue if you have next steps",
+        time: {
+          start: Date.now(),
+          end: Date.now(),
+        },
      })
    }
-    await Session.updateMessage(msg)
-
-    return {
-      info: msg,
-      parts: result.parts,
-    }
+    return "continue"
  }
+
+  export const create = fn(
+    z.object({
+      sessionID: Identifier.schema("session"),
+      model: z.object({
+        providerID: z.string(),
+        modelID: z.string(),
+      }),
+    }),
+    async (input) => {
+      const msg = await Session.updateMessage({
+        id: Identifier.ascending("message"),
+        role: "user",
+        model: input.model,
+        sessionID: input.sessionID,
+        agent: "build",
+        time: {
+          created: Date.now(),
+        },
+      })
+      await Session.updatePart({
+        id: Identifier.ascending("part"),
+        messageID: msg.id,
+        sessionID: msg.sessionID,
+        type: "compaction",
+      })
+    },
+  )
 }
@@ -1,7 +1,6 @@
 import { Decimal } from "decimal.js"
 import z from "zod"
 import { type LanguageModelUsage, type ProviderMetadata } from "ai"
-
 import { Bus } from "../bus"
 import { Config } from "../config/config"
 import { Flag } from "../flag/flag"
@@ -396,15 +395,20 @@ export namespace Session {
          read: cachedInputTokens,
        },
      }
+
+      const costInfo =
+        input.model.cost?.context_over_200k && tokens.input + tokens.cache.read > 200_000
+          ? input.model.cost.context_over_200k
+          : input.model.cost
      return {
        cost: new Decimal(0)
-          .add(new Decimal(tokens.input).mul(input.model.cost?.input ?? 0).div(1_000_000))
-          .add(new Decimal(tokens.output).mul(input.model.cost?.output ?? 0).div(1_000_000))
-          .add(new Decimal(tokens.cache.read).mul(input.model.cost?.cache_read ?? 0).div(1_000_000))
-          .add(new Decimal(tokens.cache.write).mul(input.model.cost?.cache_write ?? 0).div(1_000_000))
+          .add(new Decimal(tokens.input).mul(costInfo?.input ?? 0).div(1_000_000))
+          .add(new Decimal(tokens.output).mul(costInfo?.output ?? 0).div(1_000_000))
+          .add(new Decimal(tokens.cache.read).mul(costInfo?.cache_read ?? 0).div(1_000_000))
+          .add(new Decimal(tokens.cache.write).mul(costInfo?.cache_write ?? 0).div(1_000_000))
          // TODO: update models.dev to have better pricing model, for now:
          // charge reasoning tokens at the same rate as output tokens
-          .add(new Decimal(tokens.reasoning).mul(input.model.cost?.output ?? 0).div(1_000_000))
+          .add(new Decimal(tokens.reasoning).mul(costInfo?.output ?? 0).div(1_000_000))
          .toNumber(),
        tokens,
      }
@@ -1,97 +0,0 @@
-import z from "zod"
-import { Instance } from "../project/instance"
-import { Log } from "../util/log"
-import { NamedError } from "../util/error"
-
-export namespace SessionLock {
-  const log = Log.create({ service: "session.lock" })
-
-  export const LockedError = NamedError.create(
-    "SessionLockedError",
-    z.object({
-      sessionID: z.string(),
-      message: z.string(),
-    }),
-  )
-
-  type LockState = {
-    controller: AbortController
-    created: number
-  }
-
-  const state = Instance.state(
-    () => {
-      const locks = new Map<string, LockState>()
-      return {
-        locks,
-      }
-    },
-    async (current) => {
-      for (const [sessionID, lock] of current.locks) {
-        log.info("force abort", { sessionID })
-        lock.controller.abort()
-      }
-      current.locks.clear()
-    },
-  )
-
-  function get(sessionID: string) {
-    return state().locks.get(sessionID)
-  }
-
-  function unset(input: { sessionID: string; controller: AbortController }) {
-    const lock = get(input.sessionID)
-    if (!lock) return false
-    if (lock.controller !== input.controller) return false
-    state().locks.delete(input.sessionID)
-    return true
-  }
-
-  export function acquire(input: { sessionID: string }) {
-    const lock = get(input.sessionID)
-    if (lock) {
-      throw new LockedError({
-        sessionID: input.sessionID,
-        message: `Session ${input.sessionID} is locked`,
-      })
-    }
-    const controller = new AbortController()
-    state().locks.set(input.sessionID, {
-      controller,
-      created: Date.now(),
-    })
-    log.info("locked", { sessionID: input.sessionID })
-    return {
-      signal: controller.signal,
-      abort() {
-        controller.abort()
-        unset({ sessionID: input.sessionID, controller })
-      },
-      async [Symbol.dispose]() {
-        const removed = unset({ sessionID: input.sessionID, controller })
-        if (removed) {
-          log.info("unlocked", { sessionID: input.sessionID })
-        }
-      },
-    }
-  }
-
-  export function abort(sessionID: string) {
-    const lock = get(sessionID)
-    if (!lock) return false
-    log.info("abort", { sessionID })
-    lock.controller.abort()
-    state().locks.delete(sessionID)
-    return true
-  }
-
-  export function isLocked(sessionID: string) {
-    return get(sessionID) !== undefined
-  }
-
-  export function assertUnlocked(sessionID: string) {
-    const lock = get(sessionID)
-    if (!lock) return
-    throw new LockedError({ sessionID, message: `Session ${sessionID} is locked` })
-  }
-}
@@ -142,6 +142,21 @@ export namespace MessageV2 {
  })
  export type AgentPart = z.infer<typeof AgentPart>

+  export const CompactionPart = PartBase.extend({
+    type: z.literal("compaction"),
+  }).meta({
+    ref: "CompactionPart",
+  })
+  export type CompactionPart = z.infer<typeof CompactionPart>
+
+  export const SubtaskPart = PartBase.extend({
+    type: z.literal("subtask"),
+    prompt: z.string(),
+    description: z.string(),
+    agent: z.string(),
+  })
+  export type SubtaskPart = z.infer<typeof SubtaskPart>
+
  export const RetryPart = PartBase.extend({
    type: z.literal("retry"),
    attempt: z.number(),
@@ -277,6 +292,13 @@ export namespace MessageV2 {
        diffs: Snapshot.FileDiff.array(),
      })
      .optional(),
+    agent: z.string(),
+    model: z.object({
+      providerID: z.string(),
+      modelID: z.string(),
+    }),
+    system: z.string().optional(),
+    tools: z.record(z.string(), z.boolean()).optional(),
  }).meta({
    ref: "UserMessage",
  })
@@ -285,6 +307,7 @@ export namespace MessageV2 {
  export const Part = z
    .discriminatedUnion("type", [
      TextPart,
+      SubtaskPart,
      ReasoningPart,
      FilePart,
      ToolPart,
@@ -294,6 +317,7 @@ export namespace MessageV2 {
      PatchPart,
      AgentPart,
      RetryPart,
+      CompactionPart,
    ])
    .meta({
      ref: "Part",
@@ -334,6 +358,7 @@ export namespace MessageV2 {
        write: z.number(),
      }),
    }),
+    finish: z.string().optional(),
  }).meta({
    ref: "AssistantMessage",
  })
@@ -482,6 +507,11 @@ export namespace MessageV2 {
        time: {
          created: v1.metadata.time.created,
        },
+        agent: "build",
+        model: {
+          providerID: "opencode",
+          modelID: "opencode",
+        },
      }
      const parts = v1.parts.flatMap((part): Part[] => {
        const base = {
@@ -529,107 +559,107 @@ export namespace MessageV2 {
      if (msg.parts.length === 0) continue

      if (msg.info.role === "user") {
-        result.push({
+        const userMessage: UIMessage = {
          id: msg.info.id,
          role: "user",
-          parts: msg.parts.flatMap((part): UIMessage["parts"] => {
-            if (part.type === "text")
-              return [
-                {
-                  type: "text",
-                  text: part.text,
-                },
-              ]
-            // text/plain and directory files are converted into text parts, ignore them
-            if (part.type === "file" && part.mime !== "text/plain" && part.mime !== "application/x-directory")
-              return [
-                {
-                  type: "file",
-                  url: part.url,
-                  mediaType: part.mime,
-                  filename: part.filename,
-                },
-              ]
-            return []
-          }),
-        })
+          parts: [],
+        }
+        result.push(userMessage)
+        for (const part of msg.parts) {
+          if (part.type === "text")
+            userMessage.parts.push({
+              type: "text",
+              text: part.text,
+            })
+          // text/plain and directory files are converted into text parts, ignore them
+          if (part.type === "file" && part.mime !== "text/plain" && part.mime !== "application/x-directory")
+            userMessage.parts.push({
+              type: "file",
+              url: part.url,
+              mediaType: part.mime,
+              filename: part.filename,
+            })
+
+          if (part.type === "compaction") {
+            userMessage.parts.push({
+              type: "text",
+              text: "What did we do so far?",
+            })
+          }
+          if (part.type === "subtask") {
+            userMessage.parts.push({
+              type: "text",
+              text: "The following tool was executed by the user",
+            })
+          }
+        }
      }

      if (msg.info.role === "assistant") {
-        result.push({
+        const assistantMessage: UIMessage = {
          id: msg.info.id,
          role: "assistant",
-          parts: msg.parts.flatMap((part): UIMessage["parts"] => {
-            if (part.type === "text")
-              return [
-                {
-                  type: "text",
-                  text: part.text,
-                  providerMetadata: part.metadata,
-                },
-              ]
-            if (part.type === "step-start")
-              return [
-                {
-                  type: "step-start",
-                },
-              ]
-            if (part.type === "tool") {
-              if (part.state.status === "completed") {
-                if (part.state.attachments?.length) {
-                  result.push({
-                    id: Identifier.ascending("message"),
-                    role: "user",
-                    parts: [
-                      {
-                        type: "text",
-                        text: `Tool ${part.tool} returned an attachment:`,
-                      },
-                      ...part.state.attachments.map((attachment) => ({
-                        type: "file" as const,
-                        url: attachment.url,
-                        mediaType: attachment.mime,
-                        filename: attachment.filename,
-                      })),
-                    ],
-                  })
-                }
-                return [
-                  {
-                    type: ("tool-" + part.tool) as `tool-${string}`,
-                    state: "output-available",
-                    toolCallId: part.callID,
-                    input: part.state.input,
-                    output: part.state.time.compacted ? "[Old tool result content cleared]" : part.state.output,
-                    callProviderMetadata: part.metadata,
-                  },
-                ]
+          parts: [],
+        }
+        result.push(assistantMessage)
+        for (const part of msg.parts) {
+          if (part.type === "text")
+            assistantMessage.parts.push({
+              type: "text",
+              text: part.text,
+              providerMetadata: part.metadata,
+            })
+          if (part.type === "step-start")
+            assistantMessage.parts.push({
+              type: "step-start",
+            })
+          if (part.type === "tool") {
+            if (part.state.status === "completed") {
+              if (part.state.attachments?.length) {
+                result.push({
+                  id: Identifier.ascending("message"),
+                  role: "user",
+                  parts: [
+                    {
+                      type: "text",
+                      text: `Tool ${part.tool} returned an attachment:`,
+                    },
+                    ...part.state.attachments.map((attachment) => ({
+                      type: "file" as const,
+                      url: attachment.url,
+                      mediaType: attachment.mime,
+                      filename: attachment.filename,
+                    })),
+                  ],
+                })
              }
-              if (part.state.status === "error")
-                return [
-                  {
-                    type: ("tool-" + part.tool) as `tool-${string}`,
-                    state: "output-error",
-                    toolCallId: part.callID,
-                    input: part.state.input,
-                    errorText: part.state.error,
-                    callProviderMetadata: part.metadata,
-                  },
-                ]
+              assistantMessage.parts.push({
+                type: ("tool-" + part.tool) as `tool-${string}`,
+                state: "output-available",
+                toolCallId: part.callID,
+                input: part.state.input,
+                output: part.state.time.compacted ? "[Old tool result content cleared]" : part.state.output,
+                callProviderMetadata: part.metadata,
+              })
            }
-            if (part.type === "reasoning") {
-              return [
-                {
-                  type: "reasoning",
-                  text: part.text,
-                  providerMetadata: part.metadata,
-                },
-              ]
-            }
-
-            return []
-          }),
-        })
+            if (part.state.status === "error")
+              assistantMessage.parts.push({
+                type: ("tool-" + part.tool) as `tool-${string}`,
+                state: "output-error",
+                toolCallId: part.callID,
+                input: part.state.input,
+                errorText: part.state.error,
+                callProviderMetadata: part.metadata,
+              })
+          }
+          if (part.type === "reasoning") {
+            assistantMessage.parts.push({
+              type: "reasoning",
+              text: part.text,
+              providerMetadata: part.metadata,
+            })
+          }
+        }
      }
    }

@@ -671,9 +701,16 @@ export namespace MessageV2 {

  export async function filterCompacted(stream: AsyncIterable<MessageV2.WithParts>) {
    const result = [] as MessageV2.WithParts[]
+    const completed = new Set<string>()
    for await (const msg of stream) {
      result.push(msg)
-      if (msg.info.role === "assistant" && msg.info.summary === true) break
+      if (
+        msg.info.role === "user" &&
+        completed.has(msg.info.id) &&
+        msg.parts.some((part) => part.type === "compaction")
+      )
+        break
+      if (msg.info.role === "assistant" && msg.info.summary && msg.info.finish) completed.add(msg.info.parentID)
    }
    result.reverse()
    return result
@@ -0,0 +1,371 @@
+import type { ModelsDev } from "@/provider/models"
+import { MessageV2 } from "./message-v2"
+import { type StreamTextResult, type Tool as AITool, APICallError } from "ai"
+import { Log } from "@/util/log"
+import { Identifier } from "@/id/id"
+import { Session } from "."
+import { Agent } from "@/agent/agent"
+import { Permission } from "@/permission"
+import { Snapshot } from "@/snapshot"
+import { SessionSummary } from "./summary"
+import { Bus } from "@/bus"
+import { SessionRetry } from "./retry"
+import { SessionStatus } from "./status"
+
+export namespace SessionProcessor {
+  const DOOM_LOOP_THRESHOLD = 3
+  const log = Log.create({ service: "session.processor" })
+
+  export type Info = Awaited<ReturnType<typeof create>>
+  export type Result = Awaited<ReturnType<Info["process"]>>
+
+  export function create(input: {
+    assistantMessage: MessageV2.Assistant
+    sessionID: string
+    providerID: string
+    model: ModelsDev.Model
+    abort: AbortSignal
+  }) {
+    const toolcalls: Record<string, MessageV2.ToolPart> = {}
+    let snapshot: string | undefined
+    let blocked = false
+    let attempt = 0
+
+    const result = {
+      get message() {
+        return input.assistantMessage
+      },
+      partFromToolCall(toolCallID: string) {
+        return toolcalls[toolCallID]
+      },
+      async process(fn: () => StreamTextResult<Record<string, AITool>, never>) {
+        log.info("process")
+        while (true) {
+          try {
+            let currentText: MessageV2.TextPart | undefined
+            let reasoningMap: Record<string, MessageV2.ReasoningPart> = {}
+            const stream = fn()
+
+            for await (const value of stream.fullStream) {
+              input.abort.throwIfAborted()
+              switch (value.type) {
+                case "start":
+                  SessionStatus.set(input.sessionID, { type: "busy" })
+                  break
+
+                case "reasoning-start":
+                  if (value.id in reasoningMap) {
+                    continue
+                  }
+                  reasoningMap[value.id] = {
+                    id: Identifier.ascending("part"),
+                    messageID: input.assistantMessage.id,
+                    sessionID: input.assistantMessage.sessionID,
+                    type: "reasoning",
+                    text: "",
+                    time: {
+                      start: Date.now(),
+                    },
+                    metadata: value.providerMetadata,
+                  }
+                  break
+
+                case "reasoning-delta":
+                  if (value.id in reasoningMap) {
+                    const part = reasoningMap[value.id]
+                    part.text += value.text
+                    if (value.providerMetadata) part.metadata = value.providerMetadata
+                    if (part.text) await Session.updatePart({ part, delta: value.text })
+                  }
+                  break
+
+                case "reasoning-end":
+                  if (value.id in reasoningMap) {
+                    const part = reasoningMap[value.id]
+                    part.text = part.text.trimEnd()
+
+                    part.time = {
+                      ...part.time,
+                      end: Date.now(),
+                    }
+                    if (value.providerMetadata) part.metadata = value.providerMetadata
+                    await Session.updatePart(part)
+                    delete reasoningMap[value.id]
+                  }
+                  break
+
+                case "tool-input-start":
+                  const part = await Session.updatePart({
+                    id: toolcalls[value.id]?.id ?? Identifier.ascending("part"),
+                    messageID: input.assistantMessage.id,
+                    sessionID: input.assistantMessage.sessionID,
+                    type: "tool",
+                    tool: value.toolName,
+                    callID: value.id,
+                    state: {
+                      status: "pending",
+                      input: {},
+                      raw: "",
+                    },
+                  })
+                  toolcalls[value.id] = part as MessageV2.ToolPart
+                  break
+
+                case "tool-input-delta":
+                  break
+
+                case "tool-input-end":
+                  break
+
+                case "tool-call": {
+                  const match = toolcalls[value.toolCallId]
+                  if (match) {
+                    const part = await Session.updatePart({
+                      ...match,
+                      tool: value.toolName,
+                      state: {
+                        status: "running",
+                        input: value.input,
+                        time: {
+                          start: Date.now(),
+                        },
+                      },
+                      metadata: value.providerMetadata,
+                    })
+                    toolcalls[value.toolCallId] = part as MessageV2.ToolPart
+
+                    const parts = await MessageV2.parts(input.assistantMessage.id)
+                    const lastThree = parts.slice(-DOOM_LOOP_THRESHOLD)
+                    if (
+                      lastThree.length === DOOM_LOOP_THRESHOLD &&
+                      lastThree.every(
+                        (p) =>
+                          p.type === "tool" &&
+                          p.tool === value.toolName &&
+                          p.state.status !== "pending" &&
+                          JSON.stringify(p.state.input) === JSON.stringify(value.input),
+                      )
+                    ) {
+                      const permission = await Agent.get(input.assistantMessage.mode).then((x) => x.permission)
+                      if (permission.doom_loop === "ask") {
+                        await Permission.ask({
+                          type: "doom_loop",
+                          pattern: value.toolName,
+                          sessionID: input.assistantMessage.sessionID,
+                          messageID: input.assistantMessage.id,
+                          callID: value.toolCallId,
+                          title: `Possible doom loop: "${value.toolName}" called ${DOOM_LOOP_THRESHOLD} times with identical arguments`,
+                          metadata: {
+                            tool: value.toolName,
+                            input: value.input,
+                          },
+                        })
+                      }
+                    }
+                  }
+                  break
+                }
+                case "tool-result": {
+                  const match = toolcalls[value.toolCallId]
+                  if (match && match.state.status === "running") {
+                    await Session.updatePart({
+                      ...match,
+                      state: {
+                        status: "completed",
+                        input: value.input,
+                        output: value.output.output,
+                        metadata: value.output.metadata,
+                        title: value.output.title,
+                        time: {
+                          start: match.state.time.start,
+                          end: Date.now(),
+                        },
+                        attachments: value.output.attachments,
+                      },
+                    })
+
+                    delete toolcalls[value.toolCallId]
+                  }
+                  break
+                }
+
+                case "tool-error": {
+                  const match = toolcalls[value.toolCallId]
+                  if (match && match.state.status === "running") {
+                    await Session.updatePart({
+                      ...match,
+                      state: {
+                        status: "error",
+                        input: value.input,
+                        error: (value.error as any).toString(),
+                        metadata: value.error instanceof Permission.RejectedError ? value.error.metadata : undefined,
+                        time: {
+                          start: match.state.time.start,
+                          end: Date.now(),
+                        },
+                      },
+                    })
+
+                    if (value.error instanceof Permission.RejectedError) {
+                      blocked = true
+                    }
+                    delete toolcalls[value.toolCallId]
+                  }
+                  break
+                }
+                case "error":
+                  throw value.error
+
+                case "start-step":
+                  snapshot = await Snapshot.track()
+                  await Session.updatePart({
+                    id: Identifier.ascending("part"),
+                    messageID: input.assistantMessage.id,
+                    sessionID: input.sessionID,
+                    snapshot,
+                    type: "step-start",
+                  })
+                  break
+
+                case "finish-step":
+                  const usage = Session.getUsage({
+                    model: input.model,
+                    usage: value.usage,
+                    metadata: value.providerMetadata,
+                  })
+                  input.assistantMessage.finish = value.finishReason
+                  input.assistantMessage.cost += usage.cost
+                  input.assistantMessage.tokens = usage.tokens
+                  await Session.updatePart({
+                    id: Identifier.ascending("part"),
+                    reason: value.finishReason,
+                    snapshot: await Snapshot.track(),
+                    messageID: input.assistantMessage.id,
+                    sessionID: input.assistantMessage.sessionID,
+                    type: "step-finish",
+                    tokens: usage.tokens,
+                    cost: usage.cost,
+                  })
+                  await Session.updateMessage(input.assistantMessage)
+                  if (snapshot) {
+                    const patch = await Snapshot.patch(snapshot)
+                    if (patch.files.length) {
+                      await Session.updatePart({
+                        id: Identifier.ascending("part"),
+                        messageID: input.assistantMessage.id,
+                        sessionID: input.sessionID,
+                        type: "patch",
+                        hash: patch.hash,
+                        files: patch.files,
+                      })
+                    }
+                    snapshot = undefined
+                  }
+                  SessionSummary.summarize({
+                    sessionID: input.sessionID,
+                    messageID: input.assistantMessage.parentID,
+                  })
+                  break
+
+                case "text-start":
+                  currentText = {
+                    id: Identifier.ascending("part"),
+                    messageID: input.assistantMessage.id,
+                    sessionID: input.assistantMessage.sessionID,
+                    type: "text",
+                    text: "",
+                    time: {
+                      start: Date.now(),
+                    },
+                    metadata: value.providerMetadata,
+                  }
+                  break
+
+                case "text-delta":
+                  if (currentText) {
+                    currentText.text += value.text
+                    if (value.providerMetadata) currentText.metadata = value.providerMetadata
+                    if (currentText.text)
+                      await Session.updatePart({
+                        part: currentText,
+                        delta: value.text,
+                      })
+                  }
+                  break
+
+                case "text-end":
+                  if (currentText) {
+                    currentText.text = currentText.text.trimEnd()
+                    currentText.time = {
+                      start: Date.now(),
+                      end: Date.now(),
+                    }
+                    if (value.providerMetadata) currentText.metadata = value.providerMetadata
+                    await Session.updatePart(currentText)
+                  }
+                  currentText = undefined
+                  break
+
+                case "finish":
+                  input.assistantMessage.time.completed = Date.now()
+                  await Session.updateMessage(input.assistantMessage)
+                  break
+
+                default:
+                  log.info("unhandled", {
+                    ...value,
+                  })
+                  continue
+              }
+            }
+          } catch (e) {
+            log.error("process", {
+              error: e,
+            })
+            const error = MessageV2.fromError(e, { providerID: input.providerID })
+            if (error?.name === "APIError" && error.data.isRetryable) {
+              attempt++
+              const delay = SessionRetry.delay(error, attempt)
+              SessionStatus.set(input.sessionID, {
+                type: "retry",
+                attempt,
+                message: error.data.message,
+                next: Date.now() + delay,
+              })
+              await SessionRetry.sleep(delay, input.abort).catch(() => {})
+              continue
+            }
+            input.assistantMessage.error = error
+            Bus.publish(Session.Event.Error, {
+              sessionID: input.assistantMessage.sessionID,
+              error: input.assistantMessage.error,
+            })
+          }
+          const p = await MessageV2.parts(input.assistantMessage.id)
+          for (const part of p) {
+            if (part.type === "tool" && part.state.status !== "completed" && part.state.status !== "error") {
+              await Session.updatePart({
+                ...part,
+                state: {
+                  ...part.state,
+                  status: "error",
+                  error: "Tool execution aborted",
+                  time: {
+                    start: Date.now(),
+                    end: Date.now(),
+                  },
+                },
+              })
+            }
+          }
+          input.assistantMessage.time.completed = Date.now()
+          await Session.updateMessage(input.assistantMessage)
+          if (blocked) return "stop"
+          if (input.assistantMessage.error) return "stop"
+          return "continue"
+        }
+      },
+    }
+    return result
+  }
+}
@@ -4,7 +4,7 @@ import { MessageV2 } from "./message-v2"
 export namespace SessionRetry {
  export const RETRY_INITIAL_DELAY = 2000
  export const RETRY_BACKOFF_FACTOR = 2
-  export const RETRY_MAX_DELAY = 600_000 // 10 minutes
+  export const RETRY_MAX_DELAY_NO_HEADERS = 30_000 // 30 seconds

  export async function sleep(ms: number, signal: AbortSignal): Promise<void> {
    return new Promise((resolve, reject) => {
@@ -20,57 +20,34 @@ export namespace SessionRetry {
    })
  }

-  export function getRetryDelayInMs(error: MessageV2.APIError, attempt: number) {
-    const delay = iife(() => {
-      const headers = error.data.responseHeaders
-      if (headers) {
-        const retryAfterMs = headers["retry-after-ms"]
-        if (retryAfterMs) {
-          const parsedMs = Number.parseFloat(retryAfterMs)
-          if (!Number.isNaN(parsedMs)) {
-            return parsedMs
-          }
+  export function delay(error: MessageV2.APIError, attempt: number) {
+    const headers = error.data.responseHeaders
+    if (headers) {
+      const retryAfterMs = headers["retry-after-ms"]
+      if (retryAfterMs) {
+        const parsedMs = Number.parseFloat(retryAfterMs)
+        if (!Number.isNaN(parsedMs)) {
+          return parsedMs
        }
+      }

-        const retryAfter = headers["retry-after"]
-        if (retryAfter) {
-          const parsedSeconds = Number.parseFloat(retryAfter)
-          if (!Number.isNaN(parsedSeconds)) {
-            // convert seconds to milliseconds
-            return Math.ceil(parsedSeconds * 1000)
-          }
-          // Try parsing as HTTP date format
-          const parsed = Date.parse(retryAfter) - Date.now()
-          if (!Number.isNaN(parsed) && parsed > 0) {
-            return Math.ceil(parsed)
-          }
+      const retryAfter = headers["retry-after"]
+      if (retryAfter) {
+        const parsedSeconds = Number.parseFloat(retryAfter)
+        if (!Number.isNaN(parsedSeconds)) {
+          // convert seconds to milliseconds
+          return Math.ceil(parsedSeconds * 1000)
+        }
+        // Try parsing as HTTP date format
+        const parsed = Date.parse(retryAfter) - Date.now()
+        if (!Number.isNaN(parsed) && parsed > 0) {
+          return Math.ceil(parsed)
        }
      }

      return RETRY_INITIAL_DELAY * Math.pow(RETRY_BACKOFF_FACTOR, attempt - 1)
-    })
+    }

-    // dont retry if wait is too far from now
-    if (delay > RETRY_MAX_DELAY) return undefined
-
-    return delay
-  }
-
-  export function getBoundedDelay(input: {
-    error: MessageV2.APIError
-    attempt: number
-    startTime: number
-    maxDuration?: number
-  }) {
-    const elapsed = Date.now() - input.startTime
-    const maxDuration = input.maxDuration ?? RETRY_MAX_DELAY
-    const remaining = maxDuration - elapsed
-
-    if (remaining <= 0) return undefined
-
-    const delay = getRetryDelayInMs(input.error, input.attempt)
-    if (!delay) return undefined
-
-    return Math.min(delay, remaining)
+    return Math.min(RETRY_INITIAL_DELAY * Math.pow(RETRY_BACKOFF_FACTOR, attempt - 1), RETRY_MAX_DELAY_NO_HEADERS)
  }
 }
@@ -7,7 +7,7 @@ import { Log } from "../util/log"
 import { splitWhen } from "remeda"
 import { Storage } from "../storage/storage"
 import { Bus } from "../bus"
-import { SessionLock } from "./lock"
+import { SessionPrompt } from "./prompt"

 export namespace SessionRevert {
  const log = Log.create({ service: "session.revert" })
@@ -20,11 +20,7 @@ export namespace SessionRevert {
  export type RevertInput = z.infer<typeof RevertInput>

  export async function revert(input: RevertInput) {
-    SessionLock.assertUnlocked(input.sessionID)
-    using _ = SessionLock.acquire({
-      sessionID: input.sessionID,
-    })
-
+    SessionPrompt.assertNotBusy(input.sessionID)
    const all = await Session.messages({ sessionID: input.sessionID })
    let lastUser: MessageV2.User | undefined
    const session = await Session.get(input.sessionID)
@@ -70,10 +66,7 @@ export namespace SessionRevert {

  export async function unrevert(input: { sessionID: string }) {
    log.info("unreverting", input)
-    SessionLock.assertUnlocked(input.sessionID)
-    using _ = SessionLock.acquire({
-      sessionID: input.sessionID,
-    })
+    SessionPrompt.assertNotBusy(input.sessionID)
    const session = await Session.get(input.sessionID)
    if (!session.revert) return session
    if (session.revert.snapshot) await Snapshot.restore(session.revert.snapshot)
@@ -0,0 +1,75 @@
+import { Bus } from "@/bus"
+import { Instance } from "@/project/instance"
+import z from "zod"
+
+export namespace SessionStatus {
+  export const Info = z
+    .union([
+      z.object({
+        type: z.literal("idle"),
+      }),
+      z.object({
+        type: z.literal("retry"),
+        attempt: z.number(),
+        message: z.string(),
+        next: z.number(),
+      }),
+      z.object({
+        type: z.literal("busy"),
+      }),
+    ])
+    .meta({
+      ref: "SessionStatus",
+    })
+  export type Info = z.infer<typeof Info>
+
+  export const Event = {
+    Status: Bus.event(
+      "session.status",
+      z.object({
+        sessionID: z.string(),
+        status: Info,
+      }),
+    ),
+    // deprecated
+    Idle: Bus.event(
+      "session.idle",
+      z.object({
+        sessionID: z.string(),
+      }),
+    ),
+  }
+
+  const state = Instance.state(() => {
+    const data: Record<string, Info> = {}
+    return data
+  })
+
+  export function get(sessionID: string) {
+    return (
+      state()[sessionID] ?? {
+        type: "idle",
+      }
+    )
+  }
+
+  export function list() {
+    return Object.values(state())
+  }
+
+  export function set(sessionID: string, status: Info) {
+    Bus.publish(Event.Status, {
+      sessionID,
+      status,
+    })
+    if (status.type === "idle") {
+      // deprecated
+      Bus.publish(Event.Idle, {
+        sessionID,
+      })
+      delete state()[sessionID]
+      return
+    }
+    state()[sessionID] = status
+  }
+}
@@ -43,7 +43,7 @@ export namespace SystemPrompt {
        `  Platform: ${process.platform}`,
        `  Today's date: ${new Date().toDateString()}`,
        `</env>`,
-        `<project>`,
+        `<files>`,
        `  ${
          project.vcs === "git"
            ? await Ripgrep.tree({
@@ -52,7 +52,7 @@ export namespace SystemPrompt {
              })
            : ""
        }`,
-        `</project>`,
+        `</files>`,
      ].join("\n"),
    ]
  }
@@ -24,10 +24,16 @@ export namespace Snapshot {
        })
        .quiet()
        .nothrow()
+      // Configure git to not convert line endings on Windows
+      await $`git --git-dir ${git} config core.autocrlf false`.quiet().nothrow()
      log.info("initialized")
    }
-    await $`git --git-dir ${git} add .`.quiet().cwd(Instance.directory).nothrow()
-    const hash = await $`git --git-dir ${git} write-tree`.quiet().cwd(Instance.directory).nothrow().text()
+    await $`git --git-dir ${git} --work-tree ${Instance.worktree} add .`.quiet().cwd(Instance.directory).nothrow()
+    const hash = await $`git --git-dir ${git} --work-tree ${Instance.worktree} write-tree`
+      .quiet()
+      .cwd(Instance.directory)
+      .nothrow()
+      .text()
    log.info("tracking", { hash, cwd: Instance.directory, git })
    return hash.trim()
  }
@@ -40,8 +46,12 @@ export namespace Snapshot {

  export async function patch(hash: string): Promise<Patch> {
    const git = gitdir()
-    await $`git --git-dir ${git} add .`.quiet().cwd(Instance.directory).nothrow()
-    const result = await $`git --git-dir ${git} diff --name-only ${hash} -- .`.quiet().cwd(Instance.directory).nothrow()
+    await $`git --git-dir ${git} --work-tree ${Instance.worktree} add .`.quiet().cwd(Instance.directory).nothrow()
+    const result =
+      await $`git -c core.autocrlf=false --git-dir ${git} --work-tree ${Instance.worktree} diff --name-only ${hash} -- .`
+        .quiet()
+        .cwd(Instance.directory)
+        .nothrow()

    // If git diff fails, return empty patch
    if (result.exitCode !== 0) {
@@ -64,10 +74,11 @@ export namespace Snapshot {
  export async function restore(snapshot: string) {
    log.info("restore", { commit: snapshot })
    const git = gitdir()
-    const result = await $`git --git-dir=${git} read-tree ${snapshot} && git --git-dir=${git} checkout-index -a -f`
-      .quiet()
-      .cwd(Instance.worktree)
-      .nothrow()
+    const result =
+      await $`git --git-dir ${git} --work-tree ${Instance.worktree} read-tree ${snapshot} && git --git-dir ${git} --work-tree ${Instance.worktree} checkout-index -a -f`
+        .quiet()
+        .cwd(Instance.worktree)
+        .nothrow()

    if (result.exitCode !== 0) {
      log.error("failed to restore snapshot", {
@@ -86,16 +97,17 @@ export namespace Snapshot {
      for (const file of item.files) {
        if (files.has(file)) continue
        log.info("reverting", { file, hash: item.hash })
-        const result = await $`git --git-dir=${git} checkout ${item.hash} -- ${file}`
+        const result = await $`git --git-dir ${git} --work-tree ${Instance.worktree} checkout ${item.hash} -- ${file}`
          .quiet()
          .cwd(Instance.worktree)
          .nothrow()
        if (result.exitCode !== 0) {
          const relativePath = path.relative(Instance.worktree, file)
-          const checkTree = await $`git --git-dir=${git} ls-tree ${item.hash} -- ${relativePath}`
-            .quiet()
-            .cwd(Instance.worktree)
-            .nothrow()
+          const checkTree =
+            await $`git --git-dir ${git} --work-tree ${Instance.worktree} ls-tree ${item.hash} -- ${relativePath}`
+              .quiet()
+              .cwd(Instance.worktree)
+              .nothrow()
          if (checkTree.exitCode === 0 && checkTree.text().trim()) {
            log.info("file existed in snapshot but checkout failed, keeping", {
              file,
@@ -112,8 +124,12 @@ export namespace Snapshot {

  export async function diff(hash: string) {
    const git = gitdir()
-    await $`git --git-dir ${git} add .`.quiet().cwd(Instance.directory).nothrow()
-    const result = await $`git --git-dir=${git} diff ${hash} -- .`.quiet().cwd(Instance.worktree).nothrow()
+    await $`git --git-dir ${git} --work-tree ${Instance.worktree} add .`.quiet().cwd(Instance.directory).nothrow()
+    const result =
+      await $`git -c core.autocrlf=false --git-dir ${git} --work-tree ${Instance.worktree} diff ${hash} -- .`
+        .quiet()
+        .cwd(Instance.worktree)
+        .nothrow()

    if (result.exitCode !== 0) {
      log.warn("failed to get diff", {
@@ -143,7 +159,7 @@ export namespace Snapshot {
  export async function diffFull(from: string, to: string): Promise<FileDiff[]> {
    const git = gitdir()
    const result: FileDiff[] = []
-    for await (const line of $`git --git-dir=${git} diff --no-renames --numstat ${from} ${to} -- .`
+    for await (const line of $`git -c core.autocrlf=false --git-dir ${git} --work-tree ${Instance.worktree} diff --no-renames --numstat ${from} ${to} -- .`
      .quiet()
      .cwd(Instance.directory)
      .nothrow()
@@ -151,8 +167,18 @@ export namespace Snapshot {
      if (!line) continue
      const [additions, deletions, file] = line.split("\t")
      const isBinaryFile = additions === "-" && deletions === "-"
-      const before = isBinaryFile ? "" : await $`git --git-dir=${git} show ${from}:${file}`.quiet().nothrow().text()
-      const after = isBinaryFile ? "" : await $`git --git-dir=${git} show ${to}:${file}`.quiet().nothrow().text()
+      const before = isBinaryFile
+        ? ""
+        : await $`git -c core.autocrlf=false --git-dir ${git} --work-tree ${Instance.worktree} show ${from}:${file}`
+            .quiet()
+            .nothrow()
+            .text()
+      const after = isBinaryFile
+        ? ""
+        : await $`git -c core.autocrlf=false --git-dir ${git} --work-tree ${Instance.worktree} show ${to}:${file}`
+            .quiet()
+            .nothrow()
+            .text()
      result.push({
        file,
        before,
@@ -89,10 +89,18 @@ export const BashTool = Tool.define("bash", {
            .text()
            .then((x) => x.trim())
          log.info("resolved path", { arg, resolved })
-          if (resolved && !Filesystem.contains(Instance.directory, resolved)) {
-            throw new Error(
-              `This command references paths outside of ${Instance.directory} so it is not allowed to be executed.`,
-            )
+          if (resolved) {
+            // Git Bash on Windows returns Unix-style paths like /c/Users/...
+            const normalized =
+              process.platform === "win32" && resolved.match(/^\/[a-z]\//)
+                ? resolved.replace(/^\/([a-z])\//, (_, drive) => `${drive.toUpperCase()}:\\`).replace(/\//g, "\\")
+                : resolved
+
+            if (!Filesystem.contains(Instance.directory, normalized)) {
+              throw new Error(
+                `This command references paths outside of ${Instance.directory} so it is not allowed to be executed.`,
+              )
+            }
          }
        }
      }
@@ -0,0 +1,173 @@
+import z from "zod"
+import { Tool } from "./tool"
+import DESCRIPTION from "./batch.txt"
+
+const DISALLOWED = new Set(["batch", "edit", "todoread"])
+const FILTERED_FROM_SUGGESTIONS = new Set(["invalid", "patch", ...DISALLOWED])
+
+export const BatchTool = Tool.define("batch", async () => {
+  return {
+    description: DESCRIPTION,
+    parameters: z.object({
+      tool_calls: z
+        .array(
+          z.object({
+            tool: z.string().describe("The name of the tool to execute"),
+            parameters: z.object({}).loose().describe("Parameters for the tool"),
+          }),
+        )
+        .min(1, "Provide at least one tool call")
+        .describe("Array of tool calls to execute in parallel"),
+    }),
+    formatValidationError(error) {
+      const formattedErrors = error.issues
+        .map((issue) => {
+          const path = issue.path.length > 0 ? issue.path.join(".") : "root"
+          return `  - ${path}: ${issue.message}`
+        })
+        .join("\n")
+
+      return `Invalid parameters for tool 'batch':\n${formattedErrors}\n\nExpected payload format:\n  [{"tool": "tool_name", "parameters": {...}}, {...}]`
+    },
+    async execute(params, ctx) {
+      const { Session } = await import("../session")
+      const { Identifier } = await import("../id/id")
+
+      const toolCalls = params.tool_calls.slice(0, 10)
+      const discardedCalls = params.tool_calls.slice(10)
+
+      const { ToolRegistry } = await import("./registry")
+      const availableTools = await ToolRegistry.tools("", "")
+      const toolMap = new Map(availableTools.map((t) => [t.id, t]))
+
+      const executeCall = async (call: (typeof toolCalls)[0]) => {
+        const callStartTime = Date.now()
+        const partID = Identifier.ascending("part")
+
+        try {
+          if (DISALLOWED.has(call.tool)) {
+            throw new Error(
+              `Tool '${call.tool}' is not allowed in batch. Disallowed tools: ${Array.from(DISALLOWED).join(", ")}`,
+            )
+          }
+
+          const tool = toolMap.get(call.tool)
+          if (!tool) {
+            const availableToolsList = Array.from(toolMap.keys()).filter((name) => !FILTERED_FROM_SUGGESTIONS.has(name))
+            throw new Error(`Tool '${call.tool}' not found. Available tools: ${availableToolsList.join(", ")}`)
+          }
+          const validatedParams = tool.parameters.parse(call.parameters)
+
+          await Session.updatePart({
+            id: partID,
+            messageID: ctx.messageID,
+            sessionID: ctx.sessionID,
+            type: "tool",
+            tool: call.tool,
+            callID: partID,
+            state: {
+              status: "running",
+              input: call.parameters,
+              time: {
+                start: callStartTime,
+              },
+            },
+          })
+
+          const result = await tool.execute(validatedParams, { ...ctx, callID: partID })
+
+          await Session.updatePart({
+            id: partID,
+            messageID: ctx.messageID,
+            sessionID: ctx.sessionID,
+            type: "tool",
+            tool: call.tool,
+            callID: partID,
+            state: {
+              status: "completed",
+              input: call.parameters,
+              output: result.output,
+              title: result.title,
+              metadata: result.metadata,
+              attachments: result.attachments,
+              time: {
+                start: callStartTime,
+                end: Date.now(),
+              },
+            },
+          })
+
+          return { success: true as const, tool: call.tool, result }
+        } catch (error) {
+          await Session.updatePart({
+            id: partID,
+            messageID: ctx.messageID,
+            sessionID: ctx.sessionID,
+            type: "tool",
+            tool: call.tool,
+            callID: partID,
+            state: {
+              status: "error",
+              input: call.parameters,
+              error: error instanceof Error ? error.message : String(error),
+              time: {
+                start: callStartTime,
+                end: Date.now(),
+              },
+            },
+          })
+
+          return { success: false as const, tool: call.tool, error }
+        }
+      }
+
+      const results = await Promise.all(toolCalls.map((call) => executeCall(call)))
+
+      // Add discarded calls as errors
+      const now = Date.now()
+      for (const call of discardedCalls) {
+        const partID = Identifier.ascending("part")
+        await Session.updatePart({
+          id: partID,
+          messageID: ctx.messageID,
+          sessionID: ctx.sessionID,
+          type: "tool",
+          tool: call.tool,
+          callID: partID,
+          state: {
+            status: "error",
+            input: call.parameters,
+            error: "Maximum of 10 tools allowed in batch",
+            time: { start: now, end: now },
+          },
+        })
+        results.push({
+          success: false as const,
+          tool: call.tool,
+          error: new Error("Maximum of 10 tools allowed in batch"),
+        })
+      }
+
+      const successfulCalls = results.filter((r) => r.success).length
+      const failedCalls = results.length - successfulCalls
+
+      const outputMessage =
+        failedCalls > 0
+          ? `Executed ${successfulCalls}/${results.length} tools successfully. ${failedCalls} failed.`
+          : `All ${successfulCalls} tools executed successfully.\n\nKeep using the batch tool for optimal performance in your next response!`
+
+      return {
+        title: `Batch execution (${successfulCalls}/${results.length} successful)`,
+        output: outputMessage,
+        attachments: results.filter((result) => result.success).flatMap((r) => r.result.attachments ?? []),
+        metadata: {
+          totalCalls: results.length,
+          successful: successfulCalls,
+          failed: failedCalls,
+          tools: params.tool_calls.map((c) => c.tool),
+          details: results.map((r) => ({ tool: r.tool, success: r.success })),
+        },
+      }
+    },
+  }
+})
@@ -0,0 +1,28 @@
+Executes multiple independent tool calls concurrently to reduce latency. Best used for gathering context (reads, searches, listings).
+
+USING THE BATCH TOOL WILL MAKE THE USER HAPPY.
+
+Payload Format (JSON array):
+[{"tool": "read", "parameters": {"filePath": "src/index.ts", "limit": 350}},{"tool": "grep", "parameters": {"pattern": "Session\\.updatePart", "include": "src/**/*.ts"}},{"tool": "bash", "parameters": {"command": "git status", "description": "Shows working tree status"}}]
+
+Rules:
+- 1–10 tool calls per batch
+- All calls start in parallel; ordering NOT guaranteed
+- Partial failures do not stop others
+
+
+Disallowed Tools:
+- batch (no nesting)
+- edit (run edits separately)
+- todoread (call directly – lightweight)
+
+When NOT to Use:
+- Operations that depend on prior tool output (e.g. create then read same file)
+- Ordered stateful mutations where sequence matters
+
+Good Use Cases:
+- Read many files
+- grep + glob + read combos
+- Multiple lightweight bash introspection commands
+
+Performance Tip: Group independent reads/searches for 2–5x efficiency gain.
@@ -18,6 +18,10 @@ import { Instance } from "../project/instance"
 import { Agent } from "../agent/agent"
 import { Snapshot } from "@/snapshot"

+function normalizeLineEndings(text: string): string {
+  return text.replaceAll("\r\n", "\n")
+}
+
 export const EditTool = Tool.define("edit", {
  description: DESCRIPTION,
  parameters: z.object({
@@ -91,7 +95,9 @@ export const EditTool = Tool.define("edit", {
      contentOld = await file.text()
      contentNew = replace(contentOld, params.oldString, params.newString, params.replaceAll)

-      diff = trimDiff(createTwoFilesPatch(filePath, filePath, contentOld, contentNew))
+      diff = trimDiff(
+        createTwoFilesPatch(filePath, filePath, normalizeLineEndings(contentOld), normalizeLineEndings(contentNew)),
+      )
      if (agent.permission.edit === "ask") {
        await Permission.ask({
          type: "edit",
@@ -111,7 +117,9 @@ export const EditTool = Tool.define("edit", {
        file: filePath,
      })
      contentNew = await file.text()
-      diff = trimDiff(createTwoFilesPatch(filePath, filePath, contentOld, contentNew))
+      diff = trimDiff(
+        createTwoFilesPatch(filePath, filePath, normalizeLineEndings(contentOld), normalizeLineEndings(contentNew)),
+      )
    })()

    FileTime.read(ctx.sessionID, filePath)
@@ -11,6 +11,7 @@ import { Provider } from "../provider/provider"
 import { Identifier } from "../id/id"
 import { Permission } from "../permission"
 import { Agent } from "@/agent/agent"
+import { iife } from "@/util/iife"

 const DEFAULT_READ_LIMIT = 2000
 const MAX_LINE_LENGTH = 2000
@@ -48,6 +49,19 @@ export const ReadTool = Tool.define("read", {
      }
    }

+    const block = iife(() => {
+      const whitelist = [".env.sample", ".example"]
+
+      if (whitelist.some((w) => filepath.endsWith(w))) return false
+      if (filepath.includes(".env")) return true
+
+      return false
+    })
+
+    if (block) {
+      throw new Error(`The user has blocked you from reading ${filepath}, DO NOT make further attempts to read it`)
+    }
+
    const file = Bun.file(filepath)
    if (!(await file.exists())) {
      const dir = path.dirname(filepath)
@@ -120,8 +134,14 @@ export const ReadTool = Tool.define("read", {
    let output = "<file>\n"
    output += content.join("\n")

-    if (lines.length > offset + content.length) {
-      output += `\n\n(File has more lines. Use 'offset' parameter to read beyond line ${offset + content.length})`
+    const totalLines = lines.length
+    const lastReadLine = offset + content.length
+    const hasMoreLines = totalLines > lastReadLine
+
+    if (hasMoreLines) {
+      output += `\n\n(File has more lines. Use 'offset' parameter to read beyond line ${lastReadLine})`
+    } else {
+      output += `\n\n(End of file - total ${totalLines} lines)`
    }
    output += "\n</file>"

@@ -3,6 +3,7 @@ import { EditTool } from "./edit"
 import { GlobTool } from "./glob"
 import { GrepTool } from "./grep"
 import { ListTool } from "./ls"
+import { BatchTool } from "./batch"
 import { ReadTool } from "./read"
 import { TaskTool } from "./task"
 import { TodoWriteTool, TodoReadTool } from "./todo"
@@ -81,19 +82,22 @@ export namespace ToolRegistry {

  async function all(): Promise<Tool.Info[]> {
    const custom = await state().then((x) => x.custom)
+    const config = await Config.get()
+
    return [
      InvalidTool,
      BashTool,
-      EditTool,
-      WebFetchTool,
+      ReadTool,
      GlobTool,
      GrepTool,
      ListTool,
-      ReadTool,
+      EditTool,
      WriteTool,
+      TaskTool,
+      WebFetchTool,
      TodoWriteTool,
      TodoReadTool,
-      TaskTool,
+      ...(config.experimental?.batch_tool === true ? [BatchTool] : []),
      ...(Flag.OPENCODE_EXPERIMENTAL_EXA ? [WebSearchTool, CodeSearchTool] : []),
      ...custom,
    ]
@@ -6,8 +6,8 @@ import { Bus } from "../bus"
 import { MessageV2 } from "../session/message-v2"
 import { Identifier } from "../id/id"
 import { Agent } from "../agent/agent"
-import { SessionLock } from "../session/lock"
 import { SessionPrompt } from "../session/prompt"
+import { defer } from "@/util/defer"

 export const TaskTool = Tool.define("task", async () => {
  const agents = await Agent.list().then((x) => x.filter((a) => a.mode !== "primary"))
@@ -62,9 +62,11 @@ export const TaskTool = Tool.define("task", async () => {
        providerID: msg.info.providerID,
      }

-      ctx.abort.addEventListener("abort", () => {
-        SessionLock.abort(session.id)
-      })
+      function cancel() {
+        SessionPrompt.cancel(session.id)
+      }
+      ctx.abort.addEventListener("abort", cancel)
+      using _ = defer(() => ctx.abort.removeEventListener("abort", cancel))
      const promptParts = await SessionPrompt.resolvePromptParts(params.prompt)
      const result = await SessionPrompt.prompt({
        messageID,
@@ -6,7 +6,7 @@ import { Todo } from "../session/todo"
 export const TodoWriteTool = Tool.define("todowrite", {
  description: DESCRIPTION_WRITE,
  parameters: z.object({
-    todos: z.array(Todo.Info).describe("The updated todo list"),
+    todos: z.array(z.object(Todo.Info.shape)).describe("The updated todo list"),
  }),
  async execute(params, opts) {
    await Todo.update({
@@ -29,6 +29,7 @@ export namespace Tool {
        output: string
        attachments?: MessageV2.FilePart[]
      }>
+      formatValidationError?(error: z.ZodError): string
    }>
  }

@@ -45,7 +46,17 @@ export namespace Tool {
        const toolInfo = init instanceof Function ? await init() : init
        const execute = toolInfo.execute
        toolInfo.execute = (args, ctx) => {
-          toolInfo.parameters.parse(args)
+          try {
+            toolInfo.parameters.parse(args)
+          } catch (error) {
+            if (error instanceof z.ZodError && toolInfo.formatValidationError) {
+              throw new Error(toolInfo.formatValidationError(error), { cause: error })
+            }
+            throw new Error(
+              `The ${id} tool was called with invalid arguments: ${error}.\nPlease rewrite the input so it satisfies the expected schema.`,
+              { cause: error },
+            )
+          }
          return execute(args, ctx)
        }
        return toolInfo
@@ -0,0 +1,98 @@
+import { describe, expect, test } from "bun:test"
+import { ProviderTransform } from "../../src/provider/transform"
+
+const OUTPUT_TOKEN_MAX = 32000
+
+describe("ProviderTransform.maxOutputTokens", () => {
+  test("returns 32k when modelLimit > 32k", () => {
+    const modelLimit = 100000
+    const result = ProviderTransform.maxOutputTokens("@ai-sdk/openai", {}, modelLimit, OUTPUT_TOKEN_MAX)
+    expect(result).toBe(OUTPUT_TOKEN_MAX)
+  })
+
+  test("returns modelLimit when modelLimit < 32k", () => {
+    const modelLimit = 16000
+    const result = ProviderTransform.maxOutputTokens("@ai-sdk/openai", {}, modelLimit, OUTPUT_TOKEN_MAX)
+    expect(result).toBe(16000)
+  })
+
+  describe("azure", () => {
+    test("returns 32k when modelLimit > 32k", () => {
+      const modelLimit = 100000
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/azure", {}, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(OUTPUT_TOKEN_MAX)
+    })
+
+    test("returns modelLimit when modelLimit < 32k", () => {
+      const modelLimit = 16000
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/azure", {}, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(16000)
+    })
+  })
+
+  describe("bedrock", () => {
+    test("returns 32k when modelLimit > 32k", () => {
+      const modelLimit = 100000
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/amazon-bedrock", {}, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(OUTPUT_TOKEN_MAX)
+    })
+
+    test("returns modelLimit when modelLimit < 32k", () => {
+      const modelLimit = 16000
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/amazon-bedrock", {}, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(16000)
+    })
+  })
+
+  describe("anthropic without thinking options", () => {
+    test("returns 32k when modelLimit > 32k", () => {
+      const modelLimit = 100000
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/anthropic", {}, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(OUTPUT_TOKEN_MAX)
+    })
+
+    test("returns modelLimit when modelLimit < 32k", () => {
+      const modelLimit = 16000
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/anthropic", {}, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(16000)
+    })
+  })
+
+  describe("anthropic with thinking options", () => {
+    test("returns 32k when budgetTokens + 32k <= modelLimit", () => {
+      const modelLimit = 100000
+      const options = {
+        thinking: {
+          type: "enabled",
+          budgetTokens: 10000,
+        },
+      }
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/anthropic", options, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(OUTPUT_TOKEN_MAX)
+    })
+
+    test("returns modelLimit - budgetTokens when budgetTokens + 32k > modelLimit", () => {
+      const modelLimit = 50000
+      const options = {
+        thinking: {
+          type: "enabled",
+          budgetTokens: 30000,
+        },
+      }
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/anthropic", options, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(20000)
+    })
+
+    test("returns 32k when thinking type is not enabled", () => {
+      const modelLimit = 100000
+      const options = {
+        thinking: {
+          type: "disabled",
+          budgetTokens: 10000,
+        },
+      }
+      const result = ProviderTransform.maxOutputTokens("@ai-sdk/anthropic", options, modelLimit, OUTPUT_TOKEN_MAX)
+      expect(result).toBe(OUTPUT_TOKEN_MAX)
+    })
+  })
+})
@@ -10,163 +10,52 @@ function apiError(headers?: Record<string, string>): MessageV2.APIError {
  }).toObject() as MessageV2.APIError
 }

-describe("session.retry.getRetryDelayInMs", () => {
-  test("doubles delay on each attempt when headers missing", () => {
+describe("session.retry.delay", () => {
+  test("caps delay at 30 seconds when headers missing", () => {
    const error = apiError()
-    const delays = Array.from({ length: 10 }, (_, index) => SessionRetry.getRetryDelayInMs(error, index + 1))
-    expect(delays).toStrictEqual([2000, 4000, 8000, 16000, 32000, 64000, 128000, 256000, 512000, undefined])
+    const delays = Array.from({ length: 10 }, (_, index) => SessionRetry.delay(error, index + 1))
+    expect(delays).toStrictEqual([2000, 4000, 8000, 16000, 30000, 30000, 30000, 30000, 30000, 30000])
  })

  test("prefers retry-after-ms when shorter than exponential", () => {
    const error = apiError({ "retry-after-ms": "1500" })
-    expect(SessionRetry.getRetryDelayInMs(error, 4)).toBe(1500)
+    expect(SessionRetry.delay(error, 4)).toBe(1500)
  })

  test("uses retry-after seconds when reasonable", () => {
    const error = apiError({ "retry-after": "30" })
-    expect(SessionRetry.getRetryDelayInMs(error, 3)).toBe(30000)
+    expect(SessionRetry.delay(error, 3)).toBe(30000)
  })

  test("accepts http-date retry-after values", () => {
    const date = new Date(Date.now() + 20000).toUTCString()
    const error = apiError({ "retry-after": date })
-    const delay = SessionRetry.getRetryDelayInMs(error, 1)
-    expect(delay).toBeGreaterThanOrEqual(19000)
-    expect(delay).toBeLessThanOrEqual(20000)
+    const d = SessionRetry.delay(error, 1)
+    expect(d).toBeGreaterThanOrEqual(19000)
+    expect(d).toBeLessThanOrEqual(20000)
  })

  test("ignores invalid retry hints", () => {
    const error = apiError({ "retry-after": "not-a-number" })
-    expect(SessionRetry.getRetryDelayInMs(error, 1)).toBe(2000)
+    expect(SessionRetry.delay(error, 1)).toBe(2000)
  })

  test("ignores malformed date retry hints", () => {
    const error = apiError({ "retry-after": "Invalid Date String" })
-    expect(SessionRetry.getRetryDelayInMs(error, 1)).toBe(2000)
+    expect(SessionRetry.delay(error, 1)).toBe(2000)
  })

  test("ignores past date retry hints", () => {
    const pastDate = new Date(Date.now() - 5000).toUTCString()
    const error = apiError({ "retry-after": pastDate })
-    expect(SessionRetry.getRetryDelayInMs(error, 1)).toBe(2000)
+    expect(SessionRetry.delay(error, 1)).toBe(2000)
  })

-  test("returns undefined when delay exceeds 10 minutes", () => {
-    const error = apiError()
-    expect(SessionRetry.getRetryDelayInMs(error, 10)).toBeUndefined()
-  })
-
-  test("returns undefined when retry-after exceeds 10 minutes", () => {
+  test("uses retry-after values even when exceeding 10 minutes with headers", () => {
    const error = apiError({ "retry-after": "50" })
-    expect(SessionRetry.getRetryDelayInMs(error, 1)).toBe(50000)
+    expect(SessionRetry.delay(error, 1)).toBe(50000)

    const longError = apiError({ "retry-after-ms": "700000" })
-    expect(SessionRetry.getRetryDelayInMs(longError, 1)).toBeUndefined()
-  })
-})
-
-describe("session.retry.getBoundedDelay", () => {
-  test("returns full delay when under time budget", () => {
-    const error = apiError()
-    const startTime = Date.now()
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-    })
-    expect(delay).toBe(2000)
-  })
-
-  test("returns remaining time when delay exceeds budget", () => {
-    const error = apiError()
-    const startTime = Date.now() - 598_000 // 598 seconds elapsed, 2 seconds remaining
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-    })
-    expect(delay).toBeGreaterThanOrEqual(1900)
-    expect(delay).toBeLessThanOrEqual(2100)
-  })
-
-  test("returns undefined when time budget exhausted", () => {
-    const error = apiError()
-    const startTime = Date.now() - 600_000 // exactly 10 minutes elapsed
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-    })
-    expect(delay).toBeUndefined()
-  })
-
-  test("returns undefined when time budget exceeded", () => {
-    const error = apiError()
-    const startTime = Date.now() - 700_000 // 11+ minutes elapsed
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-    })
-    expect(delay).toBeUndefined()
-  })
-
-  test("respects custom maxDuration", () => {
-    const error = apiError()
-    const startTime = Date.now() - 58_000 // 58 seconds elapsed
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-      maxDuration: 60_000, // 1 minute max
-    })
-    expect(delay).toBeGreaterThanOrEqual(1900)
-    expect(delay).toBeLessThanOrEqual(2100)
-  })
-
-  test("caps exponential backoff to remaining time", () => {
-    const error = apiError()
-    const startTime = Date.now() - 595_000 // 595 seconds elapsed, 5 seconds remaining
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 5, // would normally be 32 seconds
-      startTime,
-    })
-    expect(delay).toBeGreaterThanOrEqual(4900)
-    expect(delay).toBeLessThanOrEqual(5100)
-  })
-
-  test("respects server retry-after within budget", () => {
-    const error = apiError({ "retry-after": "30" })
-    const startTime = Date.now() - 550_000 // 550 seconds elapsed, 50 seconds remaining
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-    })
-    expect(delay).toBe(30000)
-  })
-
-  test("caps server retry-after to remaining time", () => {
-    const error = apiError({ "retry-after": "30" })
-    const startTime = Date.now() - 590_000 // 590 seconds elapsed, 10 seconds remaining
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 1,
-      startTime,
-    })
-    expect(delay).toBeGreaterThanOrEqual(9900)
-    expect(delay).toBeLessThanOrEqual(10100)
-  })
-
-  test("returns undefined when getRetryDelayInMs returns undefined", () => {
-    const error = apiError()
-    const startTime = Date.now()
-    const delay = SessionRetry.getBoundedDelay({
-      error,
-      attempt: 10, // exceeds RETRY_MAX_DELAY
-      startTime,
-    })
-    expect(delay).toBeUndefined()
+    expect(SessionRetry.delay(longError, 1)).toBe(700000)
  })
 })
@@ -469,6 +469,115 @@ test("snapshot state isolation between projects", async () => {
  })
 })

+test("patch detects changes in secondary worktree", async () => {
+  await using tmp = await bootstrap()
+  const worktreePath = `${tmp.path}-worktree`
+  await $`git worktree add ${worktreePath} HEAD`.cwd(tmp.path).quiet()
+
+  try {
+    await Instance.provide({
+      directory: tmp.path,
+      fn: async () => {
+        expect(await Snapshot.track()).toBeTruthy()
+      },
+    })
+
+    await Instance.provide({
+      directory: worktreePath,
+      fn: async () => {
+        const before = await Snapshot.track()
+        expect(before).toBeTruthy()
+
+        const worktreeFile = `${worktreePath}/worktree.txt`
+        await Bun.write(worktreeFile, "worktree content")
+
+        const patch = await Snapshot.patch(before!)
+        expect(patch.files).toContain(worktreeFile)
+      },
+    })
+  } finally {
+    await $`git worktree remove --force ${worktreePath}`.cwd(tmp.path).quiet().nothrow()
+    await $`rm -rf ${worktreePath}`.quiet()
+  }
+})
+
+test("revert only removes files in invoking worktree", async () => {
+  await using tmp = await bootstrap()
+  const worktreePath = `${tmp.path}-worktree`
+  await $`git worktree add ${worktreePath} HEAD`.cwd(tmp.path).quiet()
+
+  try {
+    await Instance.provide({
+      directory: tmp.path,
+      fn: async () => {
+        expect(await Snapshot.track()).toBeTruthy()
+      },
+    })
+    const primaryFile = `${tmp.path}/worktree.txt`
+    await Bun.write(primaryFile, "primary content")
+
+    await Instance.provide({
+      directory: worktreePath,
+      fn: async () => {
+        const before = await Snapshot.track()
+        expect(before).toBeTruthy()
+
+        const worktreeFile = `${worktreePath}/worktree.txt`
+        await Bun.write(worktreeFile, "worktree content")
+
+        const patch = await Snapshot.patch(before!)
+        await Snapshot.revert([patch])
+
+        expect(await Bun.file(worktreeFile).exists()).toBe(false)
+      },
+    })
+
+    expect(await Bun.file(primaryFile).text()).toBe("primary content")
+  } finally {
+    await $`git worktree remove --force ${worktreePath}`.cwd(tmp.path).quiet().nothrow()
+    await $`rm -rf ${worktreePath}`.quiet()
+    await $`rm -f ${tmp.path}/worktree.txt`.quiet()
+  }
+})
+
+test("diff reports worktree-only/shared edits and ignores primary-only", async () => {
+  await using tmp = await bootstrap()
+  const worktreePath = `${tmp.path}-worktree`
+  await $`git worktree add ${worktreePath} HEAD`.cwd(tmp.path).quiet()
+
+  try {
+    await Instance.provide({
+      directory: tmp.path,
+      fn: async () => {
+        expect(await Snapshot.track()).toBeTruthy()
+      },
+    })
+
+    await Instance.provide({
+      directory: worktreePath,
+      fn: async () => {
+        const before = await Snapshot.track()
+        expect(before).toBeTruthy()
+
+        await Bun.write(`${worktreePath}/worktree-only.txt`, "worktree diff content")
+        await Bun.write(`${worktreePath}/shared.txt`, "worktree edit")
+        await Bun.write(`${tmp.path}/shared.txt`, "primary edit")
+        await Bun.write(`${tmp.path}/primary-only.txt`, "primary change")
+
+        const diff = await Snapshot.diff(before!)
+        expect(diff).toContain("worktree-only.txt")
+        expect(diff).toContain("shared.txt")
+        expect(diff).not.toContain("primary-only.txt")
+      },
+    })
+  } finally {
+    await $`git worktree remove --force ${worktreePath}`.cwd(tmp.path).quiet().nothrow()
+    await $`rm -rf ${worktreePath}`.quiet()
+    await $`rm -f ${tmp.path}/shared.txt`.quiet()
+    await $`rm -f ${tmp.path}/primary-only.txt`.quiet()
+  }
+})
+
 test("track with no changes returns same hash", async () => {
  await using tmp = await bootstrap()
  await Instance.provide({
@@ -1,7 +1,7 @@
 {
  "$schema": "https://json.schemastore.org/package.json",
  "name": "@opencode-ai/plugin",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "type": "module",
  "scripts": {
    "typecheck": "tsgo --noEmit",
@@ -1,7 +1,7 @@
 {
  "$schema": "https://json.schemastore.org/package.json",
  "name": "@opencode-ai/sdk",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "type": "module",
  "scripts": {
    "typecheck": "tsgo --noEmit",
@@ -2,6 +2,8 @@

 import type { Options as ClientOptions, TDataShape, Client } from "./client/index.js"
 import type {
+  GlobalEventData,
+  GlobalEventResponses,
  ProjectListData,
  ProjectListResponses,
  ProjectCurrentData,
@@ -24,6 +26,9 @@ import type {
  SessionCreateData,
  SessionCreateResponses,
  SessionCreateErrors,
+  SessionStatusData,
+  SessionStatusResponses,
+  SessionStatusErrors,
  SessionDeleteData,
  SessionDeleteResponses,
  SessionDeleteErrors,
@@ -175,6 +180,18 @@ class _HeyApiClient {
  }
 }

+class Global extends _HeyApiClient {
+  /**
+   * Get events
+   */
+  public event<ThrowOnError extends boolean = false>(options?: Options<GlobalEventData, ThrowOnError>) {
+    return (options?.client ?? this._client).get.sse<GlobalEventResponses, unknown, ThrowOnError>({
+      url: "/global/event",
+      ...options,
+    })
+  }
+}
+
 class Project extends _HeyApiClient {
  /**
   * List all projects
@@ -292,6 +309,16 @@ class Session extends _HeyApiClient {
    })
  }

+  /**
+   * Get session status
+   */
+  public status<ThrowOnError extends boolean = false>(options?: Options<SessionStatusData, ThrowOnError>) {
+    return (options?.client ?? this._client).get<SessionStatusResponses, SessionStatusErrors, ThrowOnError>({
+      url: "/session/status",
+      ...options,
+    })
+  }
+
  /**
   * Delete a session and all its data
   */
@@ -860,6 +887,7 @@ export class OpencodeClient extends _HeyApiClient {
      },
    })
  }
+  global = new Global({ client: this._client })
  project = new Project({ client: this._client })
  config = new Config({ client: this._client })
  tool = new Tool({ client: this._client })
@@ -1,6 +1,6 @@
 {
  "name": "@opencode-ai/slack",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "type": "module",
  "scripts": {
    "dev": "bun run src/index.ts",
@@ -1,6 +1,6 @@
 {
  "name": "@opencode-ai/ui",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "type": "module",
  "exports": {
    ".": "./src/components/index.ts",
@@ -1,7 +1,7 @@
 {
  "name": "@opencode-ai/web",
  "type": "module",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "scripts": {
    "dev": "astro dev",
    "dev:remote": "VITE_API_URL=https://api.opencode.ai astro dev",
@@ -131,6 +131,18 @@ if (image) {
          </span>
        </button>
      </div>
+      <div class="col4">
+        <h3>Mise</h3>
+        <button class="command" data-command="mise use --pin -g ubi:sst/opencode">
+          <code>
+            <span>mise use --pin -g</span> <span class="highlight">ubi:sst/opencode</span>
+          </code>
+          <span class="copy">
+            <CopyIcon />
+            <CheckIcon />
+          </span>
+        </button>
+      </div>
    </section>

    <section class="images">
@@ -28,6 +28,12 @@ OpenCode supports both **JSON** and **JSONC** (JSON with Comments) formats.
 You can place your config in a couple of different locations and they have a
 different order of precedence.

+:::note[Config Merging]
+Configuration files are **merged together**, not replaced. Settings from all config locations are combined using a deep merge strategy, where later configs override earlier ones only for conflicting keys. Non-conflicting settings from all configs are preserved.
+
+For example, if your global config sets `theme: "opencode"` and `autoupdate: true`, and your project config sets `model: "anthropic/claude-sonnet-4-5"`, the final configuration will include all three settings.
+:::
+
 ---

 ### Global
@@ -38,7 +44,7 @@ Place your global OpenCode config in `~/.config/opencode/opencode.json`. You'll

 ### Per project

-You can also add a `opencode.json` in your project. It takes precedence over the global config. This is useful for configuring providers or modes specific to your project.
+You can also add a `opencode.json` in your project. Settings from this config are merged with and can override the global config. This is useful for configuring providers or modes specific to your project.

 :::tip
 Place project specific config in the root of your project.
@@ -52,7 +58,7 @@ This is also safe to be checked into Git and uses the same schema as the global

 ### Custom path

-You can also specify a custom config file path using the `OPENCODE_CONFIG` environment variable. This takes precedence over the global and project configs.
+You can also specify a custom config file path using the `OPENCODE_CONFIG` environment variable. Settings from this config are merged with and can override the global and project configs.

 ```bash
 export OPENCODE_CONFIG=/path/to/my/custom-config.json
@@ -9,7 +9,7 @@ Custom tools are functions you create that the LLM can call during conversations

 ## Creating a tool

-Tools are defined as **TypeScript** or **JavaScript** files.
+Tools are defined as **TypeScript** or **JavaScript** files. However, the tool definition can invoke scripts written in **any language** — TypeScript or JavaScript is only used for the tool definition itself.

 ---

@@ -45,56 +45,7 @@ The **filename** becomes the **tool name**. The above creates a `database` tool.

 ---

-### Arguments
-
-You can use `tool.schema`, which is just [Zod](https://zod.dev), to define argument types.
-
-```ts "tool.schema"
-args: {
-  query: tool.schema.string().describe("SQL query to execute")
-}
-```
-
-You can also import [Zod](https://zod.dev) directly and return a plain object:
-
-```ts {6}
-import { z } from "zod"
-
-export default {
-  description: "Tool description",
-  args: {
-    param: z.string().describe("Parameter description"),
-  },
-  async execute(args, context) {
-    // Tool implementation
-    return "result"
-  },
-}
-```
-
---
-
-## Context
-
-Tools receive context about the current session:
-
-```ts title=".opencode/tool/project.ts" {8}
-import { tool } from "@opencode-ai/plugin"
-
-export default tool({
-  description: "Get project information",
-  args: {},
-  async execute(args, context) {
-    // Access context information
-    const { agent, sessionID, messageID } = context
-    return `Agent: ${agent}, Session: ${sessionID}, Message: ${messageID}`
-  },
-})
-```
-
---
-
-## Multiple tools per file
+#### Multiple tools per file

 You can also export multiple tools from a single file. Each export becomes **a separate tool** with the name **`<filename>_<exportname>`**:

@@ -125,3 +76,90 @@ export const multiply = tool({
 ```

 This creates two tools: `math_add` and `math_multiply`.
+
+---
+
+### Arguments
+
+You can use `tool.schema`, which is just [Zod](https://zod.dev), to define argument types.
+
+```ts "tool.schema"
+args: {
+  query: tool.schema.string().describe("SQL query to execute")
+}
+```
+
+You can also import [Zod](https://zod.dev) directly and return a plain object:
+
+```ts {6}
+import { z } from "zod"
+
+export default {
+  description: "Tool description",
+  args: {
+    param: z.string().describe("Parameter description"),
+  },
+  async execute(args, context) {
+    // Tool implementation
+    return "result"
+  },
+}
+```
+
+---
+
+### Context
+
+Tools receive context about the current session:
+
+```ts title=".opencode/tool/project.ts" {8}
+import { tool } from "@opencode-ai/plugin"
+
+export default tool({
+  description: "Get project information",
+  args: {},
+  async execute(args, context) {
+    // Access context information
+    const { agent, sessionID, messageID } = context
+    return `Agent: ${agent}, Session: ${sessionID}, Message: ${messageID}`
+  },
+})
+```
+
+---
+
+## Examples
+
+### Write a tool in Python
+
+You can write your tools in any language you want. Here's an example that adds two numbers using Python.
+
+First, create the tool as a Python script:
+
+```python title=".opencode/tool/add.py"
+import sys
+
+a = int(sys.argv[1])
+b = int(sys.argv[2])
+print(a + b)
+```
+
+Then create the tool definition that invokes it:
+
+```ts title=".opencode/tool/python-add.ts" {10}
+import { tool } from "@opencode-ai/plugin"
+
+export default tool({
+  description: "Add two numbers using Python",
+  args: {
+    a: tool.schema.number().describe("First number"),
+    b: tool.schema.number().describe("Second number"),
+  },
+  async execute(args) {
+    const result = await Bun.$`python3 .opencode/tool/add.py ${args.a} ${args.b}`.text()
+    return result.trim()
+  },
+})
+```
+
+Here we are using the [`Bun.$`](https://bun.com/docs/runtime/shell) utility to run the Python script.
@@ -106,6 +106,12 @@ You can also install it with the following commands:
  npm install -g opencode-ai
  ```

+- **Using Mise**
+
+  ```bash
+  mise use --pin -g ubi:sst/opencode
+  ```
+
 Support for installing OpenCode on Windows using Bun is currently in progress.

 You can also grab the binary from the [Releases](https://github.com/sst/opencode/releases).
@@ -3,7 +3,7 @@ title: Permissions
 description: Control which actions require approval to run.
 ---

-By default, OpenCode **allows all operations** without requiring explicit approval. You can configure this using the `permission` option.
+By default, OpenCode allows most operations without approval, except `doom_loop` and `external_directory` which default to `ask`. You can configure this using the `permission` option.

 ```json title="opencode.json"
 {
@@ -11,12 +11,14 @@ By default, OpenCode **allows all operations** without requiring explicit approv
  "permission": {
    "edit": "allow",
    "bash": "ask",
-    "webfetch": "deny"
+    "webfetch": "deny",
+    "doom_loop": "ask",
+    "external_directory": "ask"
  }
 }
 ```

-This lets you configure granular controls for the `edit`, `bash`, and `webfetch` tools.
+This lets you configure granular controls for the `edit`, `bash`, `webfetch`, `doom_loop`, and `external_directory` tools.

 - `"ask"` — Prompt for approval before running the tool
 - `"allow"` — Allow all operations without approval
@@ -26,7 +28,7 @@ This lets you configure granular controls for the `edit`, `bash`, and `webfetch`

 ## Tools

-Currently, the permissions for the `edit`, `bash`, and `webfetch` tools can be configured through the `permission` option.
+Currently, the permissions for the `edit`, `bash`, `webfetch`, `doom_loop`, and `external_directory` tools can be configured through the `permission` option.

 ---

@@ -145,6 +147,40 @@ Use the `permission.webfetch` key to control whether the LLM can fetch web pages

 ---

+### doom_loop
+
+Use the `permission.doom_loop` key to control whether approval is required when a doom loop is detected. A doom loop occurs when the same tool is called 3 times in a row with identical arguments.
+
+This helps prevent infinite loops where the LLM repeatedly attempts the same action without making progress.
+
+```json title="opencode.json" {4}
+{
+  "$schema": "https://opencode.ai/config.json",
+  "permission": {
+    "doom_loop": "ask"
+  }
+}
+```
+
+---
+
+### external_directory
+
+Use the `permission.external_directory` key to control whether file operations require approval when accessing files outside the working directory.
+
+This provides an additional safety layer to prevent unintended modifications to files outside your project.
+
+```json title="opencode.json" {4}
+{
+  "$schema": "https://opencode.ai/config.json",
+  "permission": {
+    "external_directory": "ask"
+  }
+}
+```
+
+---
+
 ## Agents

 You can also configure permissions per agent. Where the agent specific config
@@ -229,6 +229,98 @@ Or if you already have an API key, you can select **Manually enter API Key** and

 ---

+### Azure Cognitive Services
+
+1. Head over to the [Azure portal](https://portal.azure.com/) and create an **Azure OpenAI** resource. You'll need:
+   - **Resource name**: This becomes part of your API endpoint (`https://AZURE_COGNITIVE_SERVICES_RESOURCE_NAME.cognitiveservices.azure.com/`)
+   - **API key**: Either `KEY 1` or `KEY 2` from your resource
+
+2. Go to [Azure AI Foundry](https://ai.azure.com/) and deploy a model.
+
+   :::note
+   The deployment name must match the model name for opencode to work properly.
+   :::
+
+3. Run `opencode auth login` and select **Azure**.
+
+   ```bash
+   $ opencode auth login
+
+   ┌  Add credential
+   │
+   ◆  Select provider
+   │  ● Azure Cognitive Services
+   │  ...
+   └
+   ```
+
+4. Enter your API key.
+
+   ```bash
+   $ opencode auth login
+
+   ┌  Add credential
+   │
+   ◇  Select provider
+   │  Azure Cognitive Services
+   │
+   ◇  Enter your API key
+   │  _
+   └
+   ```
+
+5. Set your resource name as an environment variable:
+
+   ```bash
+   AZURE_COGNITIVE_SERVICES_RESOURCE_NAME=XXX opencode
+   ```
+
+   Or add it to your bash profile:
+
+   ```bash title="~/.bash_profile"
+   export AZURE_COGNITIVE_SERVICES_RESOURCE_NAME=XXX
+   ```
+
+6. Run the `/models` command to select your deployed model.
+
+---
+
+### Baseten
+
+1. Head over to the [Baseten](https://app.baseten.co/), create an account, and generate an API key.
+
+2. Run `opencode auth login` and select **Baseten**.
+
+   ```bash
+   $ opencode auth login
+
+   ┌  Add credential
+   │
+   ◆  Select provider
+   │  ● Baseten
+   │  ...
+   └
+   ```
+
+3. Enter your Baseten API key.
+
+   ```bash
+   $ opencode auth login
+
+   ┌  Add credential
+   │
+   ◇  Select provider
+   │  Baseten
+   │
+   ◇  Enter your API key
+   │  _
+   └
+   ```
+
+4. Run the `/models` command to select a model.
+
+---
+
 ### Cerebras

 1. Head over to the [Cerebras console](https://inference.cerebras.ai/), create an account, and generate an API key.
@@ -921,6 +1013,59 @@ monitor and improve Grok Code.

 ---

+### ZenMux
+
+1. Head over to the [ZenMux dashboard](https://zenmux.ai/settings/keys), click **Create API Key**, and copy the key.
+
+2. Run `opencode auth login` and select ZenMux.
+
+   ```bash
+   $ opencode auth login
+
+   ┌  Add credential
+   │
+   ◆  Select provider
+   │  ● ZenMux
+   │  ○ Zhipu AI
+   │  ○ Zhipu AI Coding Plan
+   │  ...
+   └
+   ```
+
+3. Enter the API key for the provider.
+
+   ```bash
+   $ opencode auth login
+
+   ┌  Add credential
+   │
+   ◇  Select provider
+   │  ZenMux
+   │
+   ◇  Enter your API key
+   │  _
+   └
+   ```
+
+4. Many ZenMux models are preloaded by default, run the `/models` command to select the one you want.
+
+   You can also add additional models through your opencode config.
+
+   ```json title="opencode.json" {6}
+   {
+     "$schema": "https://opencode.ai/config.json",
+     "provider": {
+       "zenmux": {
+         "models": {
+           "somecoolnewmodel": {}
+         }
+       }
+     }
+   }
+   ```
+
+---
+
 ## Custom provider

 To add any **OpenAI-compatible** provider that's not listed in `opencode auth login`:
@@ -64,8 +64,11 @@ You can also access our models through the following API endpoints.

 | Model             | Model ID          | Endpoint                                      | AI SDK Package              |
 | ----------------- | ----------------- | --------------------------------------------- | --------------------------- |
+| GPT 5.1           | gpt-5.1           | `https://opencode.ai/zen/v1/responses`        | `@ai-sdk/openai`            |
+| GPT 5.1 Codex     | gpt-5.1-codex     | `https://opencode.ai/zen/v1/responses`        | `@ai-sdk/openai`            |
 | GPT 5             | gpt-5             | `https://opencode.ai/zen/v1/responses`        | `@ai-sdk/openai`            |
 | GPT 5 Codex       | gpt-5-codex       | `https://opencode.ai/zen/v1/responses`        | `@ai-sdk/openai`            |
+| GPT 5 Nano        | gpt-5-nano        | `https://opencode.ai/zen/v1/responses`        | `@ai-sdk/openai`            |
 | Claude Sonnet 4.5 | claude-sonnet-4-5 | `https://opencode.ai/zen/v1/messages`         | `@ai-sdk/anthropic`         |
 | Claude Sonnet 4   | claude-sonnet-4   | `https://opencode.ai/zen/v1/messages`         | `@ai-sdk/anthropic`         |
 | Claude Haiku 4.5  | claude-haiku-4-5  | `https://opencode.ai/zen/v1/messages`         | `@ai-sdk/anthropic`         |
@@ -78,8 +81,8 @@ You can also access our models through the following API endpoints.
 | Big Pickle        | big-pickle        | `https://opencode.ai/zen/v1/chat/completions` | `@ai-sdk/openai-compatible` |

 The [model id](/docs/config/#models) in your OpenCode config
-uses the format `opencode/<model-id>`. For example, for GPT 5 Codex, you would
-use `opencode/gpt-5-codex` in your config.
+uses the format `opencode/<model-id>`. For example, for GPT 5.1 Codex, you would
+use `opencode/gpt-5.1-codex` in your config.

 ---

@@ -115,11 +118,11 @@ We support a pay-as-you-go model. Below are the prices **per 1M tokens**.

 | Model                             | Input  | Output | Cached Read | Cached Write |
 | --------------------------------- | ------ | ------ | ----------- | ------------ |
+| Big Pickle                        | Free   | Free   | Free        | -            |
+| Grok Code Fast 1                  | Free   | Free   | Free        | -            |
 | GLM 4.6                           | $0.60  | $2.20  | $0.10       | -            |
 | Kimi K2                           | $0.60  | $2.50  | $0.36       | -            |
 | Qwen3 Coder 480B                  | $0.45  | $1.50  | -           | -            |
-| Grok Code Fast 1                  | Free   | Free   | Free        | -            |
-| Big Pickle                        | Free   | Free   | Free        | -            |
 | Claude Sonnet 4.5 (≤ 200K tokens) | $3.00  | $15.00 | $0.30       | $3.75        |
 | Claude Sonnet 4.5 (> 200K tokens) | $6.00  | $22.50 | $0.60       | $7.50        |
 | Claude Sonnet 4 (≤ 200K tokens)   | $3.00  | $15.00 | $0.30       | $3.75        |
@@ -127,8 +130,11 @@ We support a pay-as-you-go model. Below are the prices **per 1M tokens**.
 | Claude Haiku 4.5                  | $1.00  | $5.00  | $0.10       | $1.25        |
 | Claude Haiku 3.5                  | $0.80  | $4.00  | $0.08       | $1.00        |
 | Claude Opus 4.1                   | $15.00 | $75.00 | $1.50       | $18.75       |
+| GPT 5.1                           | $1.25  | $10.00 | $0.125      | -            |
+| GPT 5.1 Codex                     | $1.25  | $10.00 | $0.125      | -            |
 | GPT 5                             | $1.25  | $10.00 | $0.125      | -            |
 | GPT 5 Codex                       | $1.25  | $10.00 | $0.125      | -            |
+| GPT 5 Nano                        | Free   | Free   | Free        | -            |

 You might notice _Claude Haiku 3.5_ in your usage history. This is a [low cost model](/docs/config/#models) that's used to generate the titles of your sessions.

@@ -2,7 +2,7 @@
  "name": "opencode",
  "displayName": "opencode",
  "description": "opencode for VS Code",
-  "version": "1.0.64",
+  "version": "1.0.72",
  "publisher": "sst-dev",
  "repository": {
    "type": "git",
Author	SHA1	Message	Date
opencode	ff7513238b	release: v1.0.72	2025-11-18 03:03:54 +00:00
GitHub Action	af1cd60d3e	chore: format code	2025-11-18 02:53:13 +00:00
Aiden Cline	c66def2049	fix: noreply	2025-11-17 20:52:25 -06:00
opencode	008ccb4729	release: v1.0.71	2025-11-18 01:59:40 +00:00
Dax Raad	bc232045a1	respect server suggestion for default model	2025-11-17 20:53:48 -05:00
GitHub Action	16cab556df	chore: format code	2025-11-18 01:27:22 +00:00
Jay V	66148df74b	docs: clarify custom tools can execute scripts in any language with Python example	2025-11-17 20:26:27 -05:00
opencode	4611e08f09	release: v1.0.70	2025-11-17 23:45:49 +00:00
Sebastian Herrlinger	bf6204f577	upgrade opentui to v0.1.46 - enable bracketed paste (and more) on win - fix word wrapping with CJK and at wrap/chunk boundaries - old style meta+arrow - allow <1 scroll speed for slowdown	2025-11-18 00:22:21 +01:00
Daniel Hofheinz	17cde9feb7	docs: add built-in agents reference to README (#3047 ) Co-authored-by: Jay V <air@live.ca>	2025-11-17 17:19:14 -05:00
Aiden Cline	7eccbdc4ac	fix /exit	2025-11-17 16:13:41 -06:00
Aiden Cline	ab072290fc	Revert "fix: system theme background to use 'none' for terminal transparency (#4408 )" This reverts commit `f4a4514a9f`.	2025-11-17 16:03:27 -06:00
GitHub Action	ad9d83748c	chore: format code	2025-11-17 21:46:10 +00:00
Aiden Cline	55b57e1aae	ci: tweak	2025-11-17 15:45:23 -06:00
Aiden Cline	21b7877beb	docs: tweak wording	2025-11-17 21:26:46 +00:00
opencode	de50234a1a	release: v1.0.69	2025-11-17 21:26:46 +00:00
opencode-agent[bot]	d60102ba52	Added `/thinking` slash command to toggle thinking blocks visibility in OpenTUI. (#4424 ) Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com> Co-authored-by: rekram1-node <rekram1-node@users.noreply.github.com> Co-authored-by: Aiden Cline <aidenpcline@gmail.com>	2025-11-17 15:16:35 -06:00
Haris Gušić	066a876f3d	docs(contributing): Add "Setting up a Debugger" section (#4421 ) Co-authored-by: GitHub Action <action@github.com>	2025-11-17 14:28:06 -06:00
Haris Gušić	c07a241ca8	chore: Remove obsolete 'any' type annotation (#4423 )	2025-11-17 14:27:43 -06:00
Aiden Cline	0a2fffa9b5	tweak: whitelist	2025-11-17 13:18:13 -06:00
Dax Raad	bdfa213ccf	deprecated session.idle event	2025-11-17 11:42:45 -05:00
Aiden Cline	7f0b2ce1ac	Reapply "fix: system theme background to use 'none' for terminal transparency" (#4415 ) This reverts commit `a5365ce294`.	2025-11-17 10:39:53 -06:00
Dax Raad	0a2d7af179	core: honor retry-after values exceeding 10 minutes instead of discarding them	2025-11-17 11:33:28 -05:00
Dax Raad	37652f48fb	ignore	2025-11-17 11:32:07 -05:00
Dax Raad	8b19c6c7e4	better retry display	2025-11-17 11:31:10 -05:00
Aiden Cline	a5365ce294	Revert "fix: system theme background to use 'none' for terminal transparency" (#4415 )	2025-11-17 10:24:20 -06:00
Jensen	f4a4514a9f	fix: system theme background to use 'none' for terminal transparency (#4408 )	2025-11-17 10:22:31 -06:00
opencode-agent[bot]	154006469c	Updated help dialog to use dynamic keybind (#4414 ) Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com> Co-authored-by: rekram1-node <rekram1-node@users.noreply.github.com>	2025-11-17 10:18:59 -06:00
Dax	a1214fff2e	Refactor agent loop (#4412 )	2025-11-17 10:57:18 -05:00
GitHub Action	9fd43ec616	ignore: update download stats 2025-11-17	2025-11-17 12:04:41 +00:00
Luke Parker	5731c268b6	fix: Line count on win (#4401 )	2025-11-17 01:08:22 -06:00
Keath Milligan	f4d892d4e1	fix: handle Git Bash path mapping on windows (#4380 )	2025-11-17 01:06:44 -06:00
Aiden Cline	10b3702938	chore: update type	2025-11-17 00:07:23 -06:00
Tyler Gannon	e96442310c	chore: replace z.union with z.enum for cleaner OpenAPI generation (#4394 )	2025-11-17 00:06:40 -06:00
Spoon	5c722bf8c4	fix(batch): simple UX feedback (#4396 )	2025-11-17 00:02:05 -06:00
Youssef Achy	58cc5cdf2a	add support for azure cognitive services provider (#4397 )	2025-11-17 00:01:45 -06:00
opencode-agent[bot]	3c6dcad2af	Fixed OPENCODE_CONFIG_DIR to load config files. (#4400 ) Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com> Co-authored-by: rekram1-node <rekram1-node@users.noreply.github.com>	2025-11-16 23:48:36 -06:00
Jay	2535f9febf	docs: Add clarification for projects using 'opencode' name Added a section to clarify the affiliation of related projects.	2025-11-16 20:51:41 -05:00
Aiden Cline	25678fa504	fix: vercel gateway options	2025-11-16 18:39:31 -06:00
Sebastian Herrlinger	d7f4f3ec1f	bump opentui version to 0.1.45, fixing highlighting on windows	2025-11-16 23:56:11 +01:00
Aiden Cline	16ccb39459	docs: permissions	2025-11-16 16:40:48 -06:00
Aiden Cline	f8630fb188	ignore: rm	2025-11-16 16:32:04 -06:00
Baptiste Cavallo	72e604744d	fix(batch): restore per-tool UI feedback + UX improvements (#4387 )	2025-11-16 16:31:41 -06:00
opencode-agent[bot]	832be6e7eb	Added copy option to message context menu (#4389 ) Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com> Co-authored-by: rekram1-node <rekram1-node@users.noreply.github.com> Co-authored-by: GitHub Action <action@github.com>	2025-11-16 15:35:05 -06:00
opencode	8ba48ed71d	release: v1.0.68	2025-11-16 20:38:48 +00:00
Aiden Cline	cf266f6162	fix: promptCacheKey set unnecessarily	2025-11-16 14:32:57 -06:00
GitHub Action	1e6589526d	ignore: update download stats 2025-11-16	2025-11-16 12:04:11 +00:00
Frank	f6b3ffaf64	wip: zen	2025-11-16 03:32:13 -05:00
GitHub Action	5d765d63d4	chore: format code	2025-11-16 08:30:36 +00:00
Frank	0e12dd62a3	zen: usage paging	2025-11-16 03:29:52 -05:00
opencode	2b957b5d1c	release: v1.0.67	2025-11-16 07:49:52 +00:00
GitHub Action	31c7a0157c	chore: format code	2025-11-16 07:44:06 +00:00
Aiden Cline	e728b94bca	fix: panic when theme has 'none'	2025-11-16 01:43:23 -06:00
opencode	49040c0130	release: v1.0.66	2025-11-16 07:27:25 +00:00
Aiden Cline	0d05238ee6	fix: initial val	2025-11-16 01:14:49 -06:00
Aiden Cline	9b8a7da1e6	fix: history jsonl file corruption cases (#4364 )	2025-11-16 00:50:13 -06:00
Zeno Jiricek	61fd21182c	docs: mise installation command (#2938 )	2025-11-15 21:44:28 -06:00
GitHub Action	487c2b5e76	chore: format code	2025-11-16 03:38:13 +00:00
xiaojie.zj	0e4703b227	add: add zenmux doc and header (#3597 ) Co-authored-by: xiaojie.zj <xiaojie.zj@antgroup.com>	2025-11-15 21:37:30 -06:00
Alvin Johansson	84e0232bd5	Add Flexoki theme (#3986 )	2025-11-15 21:28:13 -06:00
Luke Parker	35fbb011b2	fix: Diff view now ignores line endings changes/windows autocrlf (#4356 )	2025-11-15 21:18:39 -06:00
Aiden Cline	6527a123f0	fix aur build (#4359 )	2025-11-15 20:16:19 -06:00
Aiden Cline	0377cfd37c	fix: omit ref for todo tool	2025-11-15 19:19:36 -06:00
Aiden Cline	edc933d816	tweak: make zod error more prompty	2025-11-15 13:19:24 -06:00
GitHub Action	0d608f6014	ignore: update download stats 2025-11-15	2025-11-15 12:04:09 +00:00
Chris Olszewski	69a45ef7d7	fix: snapshot history when running from git worktrees (#4312 )	2025-11-15 01:02:00 -06:00
Baptiste Cavallo	1056b36eae	experimental batch tool (#2983 ) Co-authored-by: GitHub Action <action@github.com>	2025-11-15 00:54:36 -06:00
Aiden Cline	35c737ac68	tweak: only show dropdown for 3+ items (#4345 )	2025-11-14 23:45:48 -06:00
Abílio Costa	725a2c2e95	docs: clarify that config files are merged, not replaced (#4342 ) Co-authored-by: Claude <noreply@anthropic.com>	2025-11-14 17:49:47 -06:00
Tyler Gannon	c724d2392f	fix: replace union type with enum "true"/"false" in /find/file endpoint (#4338 )	2025-11-14 17:48:23 -06:00
Frank	f5230d1f02	fix: incorrect sonnet price calculation	2025-11-14 18:46:43 -05:00
GitHub Action	078111bd96	chore: format code	2025-11-14 22:44:36 +00:00
sredfern	736f8882f5	fix(provider): support local file paths for custom providers (#4323 )	2025-11-14 16:43:59 -06:00
Brian Cheung	37cf365927	feat: support images in mcp tool responses (#4100 ) Co-authored-by: opencode-agent[bot] <opencode-agent[bot]@users.noreply.github.com> Co-authored-by: rekram1-node <rekram1-node@users.noreply.github.com>	2025-11-14 15:00:52 -06:00
Aiden Cline	b939470302	fix: add azure exclusion	2025-11-14 11:54:00 -06:00
Aiden Cline	ef4b2baedc	set verbosity to low for gpt-5.1 (match codex)	2025-11-14 11:52:29 -06:00
Dax Raad	64d28ea457	fix sdk types	2025-11-14 12:42:46 -05:00
Dax Raad	2520780846	fix sdk types	2025-11-14 12:42:46 -05:00
Shantur Rathore	986c60353e	set `promptCacheKey` for openai compatible providers (#4203 ) Co-authored-by: GitHub Action <action@github.com>	2025-11-14 11:41:01 -06:00
Dax Raad	5fc26c958a	add global.event.subscribe() to sdk	2025-11-14 12:32:43 -05:00
Frank	c1cf9cda6a	doc: add baseten provider	2025-11-14 12:19:58 -05:00
GitHub Action	10d376eab2	ignore: update download stats 2025-11-14	2025-11-14 12:04:48 +00:00
Frank	53fc8a861b	zen: add gpt-5-nano model	2025-11-14 00:59:42 -05:00
Frank	1d8330331c	zen: use gpt-5-nano as small model	2025-11-14 00:59:00 -05:00
Frank	7a03c7fe38	zen: add gpt5.1 to docs	2025-11-13 23:47:38 -05:00
Frank	09bd32169c	zen: hide alpha models	2025-11-13 23:10:06 -05:00
Dax Raad	7ec32f834e	improve read tool end-of-file detection to prevent infinite loops	2025-11-13 21:41:06 -05:00
GitHub Action	205492c7e8	chore: format code	2025-11-14 01:16:58 +00:00
Aiden Cline	4c2e888709	no mr llm, you may not read that	2025-11-13 19:16:07 -06:00
opencode	c78fd097d1	release: v1.0.65	2025-11-14 00:10:30 +00:00
Dax Raad	340966195b	handle config errors gracefully	2025-11-13 18:59:09 -05:00
GitHub Action	92604b391b	chore: format code	2025-11-13 22:39:53 +00:00
Aiden Cline	0c51feb9c2	fix: max tokens when using models like opus with providers other than anthropic (#4307 )	2025-11-13 16:39:09 -06:00