Merge origin/main (7719d348ca) into the 0.10.2 candidate

origin/main adds only maintainer docs and skills (AGENTS.md, docs/skills/*).
No code changes, no conflicts with the candidate.

AGENTS.md

@@ -48,7 +48,7 @@ base prompt". Two more corollaries earned here:
 - Before adding a module named `model_*`, `*_config`, `provider_*`, or
   anything that "bridges", "mirrors", or "stages" an existing thing, grep
   for the existing thing and edit it. A new layer must name the predecessor
-  it replaces in the module doc; otherwise edit the original.
+  it replaces in its commit message; otherwise edit the original.
 - Prefer the simplest implementation that preserves observable contracts. A
   rewrite is acceptable when justified by product intent and observed behavior,
   not as a shortcut around understanding existing code.
@@ -62,6 +62,11 @@ base prompt". Two more corollaries earned here:
   A fresh worktree is for conflicting, dirty, stale, or independent lanes
   (see `cw-land`), not for parallel agents on the same lane. Local commit
   permission never implies push, merge, tag, release, or deploy permission.
+- **Maintainers do not open PR branches** (founder, 2026-10-08: more PRs means
+  more CI and slower work). Verified slices land directly on `main` and are
+  pushed: run the focused gate, put the real counts in the commit message,
+  fetch and rebase onto `origin/main` and retry on rejection. Never force-push
+  or rewrite a shared ref. Contributor PRs still merge as PRs.
 - When the task is local-only, stay fully offline: no browsing, GitHub or remote
   Git operations, downloads, dependency installation, provider calls, or
   source/diff transmission. Record the missing external receipt and keep working
@@ -145,7 +150,7 @@ casually it reads as taking the work even when credit is preserved.
 
 - The model-facing subagent tool is `agent`; `agent_open`/`agent_eval`/
   `agent_close`/`delegate_to_agent` are removed surfaces. If the shape must
-  move, move the code and add the guard test that judges the new shape.
+  move, move the code and its existing guard test with it.
 - `BASE_PROMPT` in `crates/tui/src/prompts/text.rs` is the sole base prompt
   by convention. Same rule: move the code, not the prose, if that changes.
 - There is exactly one turn loop: `Engine::run_turn` in
@@ -178,18 +183,17 @@ casually it reads as taking the work even when credit is preserved.
 - Product intent and observed runtime behavior outrank a test's preferred
   implementation shape. Fix the product; do not contort production code to
   preserve a brittle assertion.
-- Code first, then tests. Write the implementation and prove it runs, then add
-  or adjust tests to cover what was actually built. Never write tests first and
-  never practice TDD here — this overrides any skill or default that mandates
-  it, including superpowers `test-driven-development`. Tests stay the gate
-  before a push; they are not the design driver. An existing test that only
-  encodes old behavior is evidence, not a veto: change it with the code rather
-  than bending the code to keep it green. This does not relax the rule under
-  "Claiming a test passed" — a regression test written *after* the fix still
-  has to be shown failing without it.
-- Tests are selective evidence, not the specification. Do not add tests by
-  default. Add or retain one when it cheaply protects a high-risk behavior such
-  as safety, data integrity, protocol compatibility, or a reproduced regression.
+- **No new tests and no new code comments unless explicitly asked**
+  (founder, 2026-10-08). Do not write new unit or integration tests unless the
+  user explicitly instructs it for that task; prove a change by running it and
+  the existing checks. Never write tests first and never practice TDD — this
+  overrides any skill or default that mandates it, including superpowers
+  `test-driven-development`. Existing tests stay the gate before a push: when
+  a change makes one wrong, update or delete it in the same slice rather than
+  bending the code to keep it green. When a regression test is requested, the
+  rule under "Claiming a test passed" still applies. Do not add new code
+  comments; the reasoning goes in the commit message, and an existing comment
+  the change makes false is fixed or removed.
 - Rewrite or remove tests that duplicate coverage, freeze internals, overspecify
   copy or layout, preserve obsolete behavior, or cost more than the risk they
   cover. Never weaken real safety or data-integrity behavior merely to make a

docs/skills/README.md

@@ -14,6 +14,17 @@ To activate:
   `/skills`, or bundle into `crates/tui/assets/skills/` + register in
   `crates/tui/src/skills/system.rs` to ship it.
 
-Skills: gh-file-issue, gh-compile-issues, gh-assign-issues, gh-find-prs,
-gh-treasure-hunt, gh-close-issues, gh-credit-harvest, codew-release-qa-sweep,
-contributor-onboarding, feedback.
+The maintainer loop, in order — each skill links the next:
+
+| Stage | Skill | Use it for |
+| --- | --- | --- |
+| 1 | [cw-orient](cw-orient/SKILL.md) | Live checkout, branch, dirt and version truth before editing |
+| 2 | [cw-slice](cw-slice/SKILL.md) | Find the existing owner, bound one slice, fix the evidence bar |
+| 3 | [cw-gates](cw-gates/SKILL.md) | The focused-to-broad verification ladder and CI budget checks |
+| 4 | [cw-dogfood](cw-dogfood/SKILL.md) | Stamped build, atomic install, real-product QA |
+| 5 | [cw-land](cw-land/SKILL.md) | Commits, PRs, contributor credit, merging under a gate |
+| 6 | [cw-handoff](cw-handoff/SKILL.md) | A paste-ready continuation grounded in live state |
+
+GitHub stewardship and release: gh-file-issue, gh-compile-issues,
+gh-assign-issues, gh-find-prs, gh-treasure-hunt, gh-close-issues,
+gh-credit-harvest, codew-release-qa-sweep, contributor-onboarding, feedback.

docs/skills/cw-gates/SKILL.md

@@ -70,7 +70,7 @@ python3 scripts/check-persistence-backlog-budget.py
 python3 scripts/check-provider-registry.py         # provider registry drift
 python3 scripts/check-command-crate-boundaries.py  # command-contract boundary
 python3 scripts/check-command-migration-manifest.py
-python3 scripts/check-tui-locale-parity.py         # touched crates/tui/locales/
+python3 scripts/check-tui-locale-parity.py         # touched crates/localization/locales/
 sh scripts/check-tui-product-vocabulary.sh
 python3 scripts/check-readme-translations.py       # touched README*.md
 ./scripts/release/check-versions.sh                # touched a version anywhere
@@ -104,14 +104,13 @@ cd web && npm ci && npm test && npm run check
 
 ## Claiming a test passed
 
-- Quote the real `test result: N passed; M failed` line, and confirm `N > 0`
-  **for the tests that cover your change**. `cargo test <filter>` exits 0 having
-  run zero tests when the filter matches nothing; an exit code alone has already
-  been mistaken for a pass here.
-- Prefer proving a regression test fails without the fix. A test that passes
-  either way pins the implementation, not the defect.
-- Audit any hand-rolled scorer before trusting its score: quote the counts it
-  actually evaluated, not the verdict line alone.
+`AGENTS.md` ("Claiming a test passed") owns the rules: quote the real count
+line with `N > 0` for the tests that cover the change, prefer a regression
+test that fails without the fix, and audit hand-rolled scorers. Two additions
+for gate runs:
+
+- Check the count **per filter group**. A green run where one required filter
+  matched zero tests is a miss, not a pass.
 - A focused rerun of a failing test distinguishes flake from regression in
   seconds. Do that before calling anything a flake, and root-cause anything that
   fails outside a known-flaky name — check for unisolated config-path reads or

docs/skills/cw-land/SKILL.md

@@ -17,7 +17,7 @@ Stage 5 of the loop: [cw-orient](../cw-orient/SKILL.md) →
 
 ## When to use
 
-- The change is verified and needs to become a commit, branch, or PR.
+- The change is verified and needs to become a commit on `main`.
 - You are landing someone else's PR, harvesting a contributor's work, or
   resolving a conflict caused by `main` moving.
 - You are about to merge something behind a required gate.
@@ -25,16 +25,20 @@ Stage 5 of the loop: [cw-orient](../cw-orient/SKILL.md) →
 ## Workflow
 
 1. **Choose the landing shape.**
-   - **Direct to `main`** is permitted for a small coherent change when this
-     checkout is current, clean, and owns the affected files. Local commit
-     permission never implies push, merge, tag, release, or deploy permission.
-   - **A worktree** is the right safety boundary for conflicting, dirty, stale,
-     or independent work — and for anything that would otherwise fight the dirt
-     you found in [cw-orient](../cw-orient/SKILL.md).
-   - **An integration branch** — `integration/<topic>-<pr>-<date>` — is the
-     normal path for anything with conflicts or several moving PRs. It is
-     cheaper than rebasing onto a `main` that keeps moving, and it leaves the
-     contributor's branch untouched.
+   - **The checkout that already exists** is the default. Direct to `main` is
+     permitted for a small coherent change when that checkout is current and
+     owns the affected files; when several agents share it, partition by file,
+     stage only your paths, and retry a commit that fails on `index.lock`.
+     Local commit permission never implies push, merge, tag, release, or
+     deploy permission.
+   - **No maintainer PR branches** (`AGENTS.md`): verified slices land on
+     `main` directly and hosted CI runs there. PRs are for contributors.
+   - **A worktree** is for a lane that genuinely conflicts with the dirt you
+     found in [cw-orient](../cw-orient/SKILL.md), or an experiment you may
+     throw away — not for parallel agents on the same lane.
+   - **An integration branch** (`integration/<topic>-<pr>-<date>`) is for
+     resolving a contributor PR that conflicts with a moving `main` when you
+     cannot push to their fork. It leaves their branch untouched.
 
 2. **Commit narrow and build-green.** One coherent change per commit; the tree
    builds at every commit. Put the real verification in the message — actual
@@ -91,10 +95,10 @@ Stage 5 of the loop: [cw-orient](../cw-orient/SKILL.md) →
    the artifact must literally say PASS, the review thread outranks the check
    rollup, and ambiguity blocks the merge — never the reverse.
 
-7. **Clean up your own lane.** When a worktree's branch lands on `main`, remove
-   the worktree (`git worktree remove <path>`). `scripts/worktree-gc.sh`
-   lists lanes by merged/dirty state; `--remove-merged --yes` prunes the
-   merged-and-clean ones. Worktree sprawl was a 560 GB problem here once.
+7. **Clean up your own lane.** When a worktree you created lands on `main`,
+   remove it (`git worktree remove <path>`) and delete its merged branch.
+   Leave other writers' worktrees alone; `git worktree list` shows them.
+   Worktree sprawl was a 560 GB problem here once.
 
 ## Red flags / don't
 

docs/skills/cw-orient/SKILL.md

@@ -38,8 +38,9 @@ This is stage 1 of the loop: **orient → [cw-slice](../cw-slice/SKILL.md) →
    git worktree list
    git branch --sort=-committerdate --format='%(committerdate:short) %(refname:short)' | head -20
    ```
-   If the work you were asked to do already has a worktree, use it. Do not
-   start a second copy of the same lane.
+   If the work you were asked to do already has a checkout or worktree, use
+   it. Do not start a second copy of the same lane, and do not create a
+   worktree just because another agent is in this one — partition by file.
 
 3. **Read the dirt before you touch it.** Modified files you did not write
    belong to someone else — another agent, another lane, an in-flight slice:
@@ -48,8 +49,8 @@ This is stage 1 of the loop: **orient → [cw-slice](../cw-slice/SKILL.md) →
    git diff --stat
    ```
    Preserve them. Leave them unstaged, and do not `git checkout --` or stash
-   another writer's work to get a clean tree. If your change genuinely conflicts
-   with the dirt, work in a fresh worktree instead.
+   another writer's work to get a clean tree. Only if your change genuinely
+   conflicts with that dirt is a fresh worktree warranted ([cw-land](../cw-land/SKILL.md)).
 
 4. **Fix which guidance applies.** The nearest scoped `AGENTS.md` wins over the
    root one, and it is where the per-area rules actually live:

docs/skills/cw-slice/SKILL.md

@@ -39,7 +39,7 @@ Stage 2 of the loop: [cw-orient](../cw-orient/SKILL.md) → **slice** →
    grep -rln 'model_\|_config\|provider_' crates/*/src --include='*.rs' | head -30
    ```
    Search behavior and symbols, not just filenames. If you find an owner, edit
-   it. If you are still adding a new layer, its module doc must name the
+   it. If you are still adding a new layer, its commit message must name the
    predecessor it replaces — otherwise you are editing the wrong file.
 
 3. **Check the contracts you are about to walk into.** One of these is
@@ -66,7 +66,7 @@ Stage 2 of the loop: [cw-orient](../cw-orient/SKILL.md) → **slice** →
 4. **Read the scoped guidance for the files you will touch.** `crates/tui/AGENTS.md`
    owns the UI contracts (one owner per fact, `codewhale_palette::grammar` semantics, typed
    state enums, toast routing, `tr(locale, MessageId::...)` for user prose).
-   `crates/tui/locales/AGENTS.md` owns string changes. `web/AGENTS.md` owns the
+   `crates/localization/locales/AGENTS.md` owns string changes. `web/AGENTS.md` owns the
    site. `docs/MOTION_CONTRACT.md` owns motion. Design law lives in
    `docs/design/`, not in a prototype file someone left in a sibling directory.
 
@@ -83,21 +83,20 @@ Stage 2 of the loop: [cw-orient](../cw-orient/SKILL.md) → **slice** →
    - whether it is cross-cutting enough to need the full sweep in
      [cw-gates](../cw-gates/SKILL.md).
 
-7. **Write the implementation first.** Code first, then tests (see `AGENTS.md`
-   — this repo does not practice TDD). Build it, prove it runs, then add or
-   adjust tests to cover what you actually built.
+7. **Write the implementation, then prove it runs.** No new tests or code
+   comments unless explicitly asked (`AGENTS.md`); update or delete existing
+   tests the change makes wrong.
 
 ## Red flags / don't
 
 - Don't add a module that "bridges", "mirrors", "stages", or "wraps" something
-  that already exists without naming that thing in the module doc.
+  that already exists without naming that thing in the commit message.
 - Don't fork a singleton (turn loop, base prompt, delegation axis,
   lifecycle system) without moving its guard test and consumers with it.
   The repo has one of each on purpose; a silent second one is the failure
   mode, not the refactor.
-- Don't write tests first. Don't add tests by default either — add one when it
-  cheaply protects safety, data integrity, protocol compatibility, or a
-  reproduced regression.
+- Don't write new tests unless explicitly asked, and never first. Don't add
+  new code comments.
 - Don't contort production code to keep a brittle assertion green. A test that
   only encodes old behavior is evidence, not a veto: change it with the code.
 - Don't cut trust-boundary validation, data-loss handling, security, or