langchain-ai/deepagents
1
0
Fork 0
mirror of https://github.com/langchain-ai/deepagents.git synced 2026-07-25 12:55:53 -04:00
Code Issues 169 Packages Projects Releases 16 Wiki Activity

[PR #1154] feat(sdk): add per-command timeout override to execute() #1175

New Issue
Open
opened 2026-02-16 09:18:20 -05:00 by yindo · 0 comments
yindo commented 2026-02-16 09:18:20 -05:00
Owner
Copy Link
Copy Source

📋 Pull Request Information

Original PR: https://github.com/langchain-ai/deepagents/pull/1154
Author: @mdrxy
Created: 2/6/2026
Status: 🔄 Open

Base: main ← Head: mdrxy/execute-timeout

📝 Commits (3)

  • e98607b feat(sdk): add per-command timeout override to execute()
  • 5b6cc15 Merge remote-tracking branch 'origin/main' into mdrxy/execute-timeout
  • 41c7d28 add 1hr max timeout

📊 Changes

21 files changed (+521 additions, -55 deletions)

View changed files

📝 libs/cli/deepagents_cli/backends.py (+15 -3)
📝 libs/cli/deepagents_cli/integrations/daytona.py (+10 -1)
📝 libs/cli/deepagents_cli/integrations/langsmith.py (+6 -2)
📝 libs/cli/deepagents_cli/integrations/modal.py (+10 -1)
📝 libs/cli/deepagents_cli/integrations/runloop.py (+7 -1)
📝 libs/cli/tests/unit_tests/test_backend_timeout.py (+1 -1)
📝 libs/deepagents/deepagents/backends/__init__.py (+2 -1)
📝 libs/deepagents/deepagents/backends/composite.py (+21 -12)
📝 libs/deepagents/deepagents/backends/local_shell.py (+44 -7)
📝 libs/deepagents/deepagents/backends/protocol.py (+13 -1)
📝 libs/deepagents/deepagents/backends/sandbox.py (+5 -0)
📝 libs/deepagents/deepagents/middleware/filesystem.py (+39 -2)
📝 libs/deepagents/tests/integration_tests/test_filesystem_middleware.py (+3 -3)
📝 libs/deepagents/tests/unit_tests/backends/test_composite_backend_async.py (+2 -2)
📝 libs/deepagents/tests/unit_tests/backends/test_sandbox_backend.py (+1 -1)
📝 libs/deepagents/tests/unit_tests/test_local_sandbox_operations.py (+7 -3)
➕ libs/deepagents/tests/unit_tests/test_local_shell.py (+102 -0)
📝 libs/deepagents/tests/unit_tests/test_middleware.py (+199 -4)
📝 libs/deepagents/tests/unit_tests/test_middleware_async.py (+6 -6)
📝 libs/harbor/deepagents_harbor/backend.py (+13 -2)

...and 1 more files

📄 Description

Prior to #1107, the CLI used a standalone ShellMiddleware (in libs/cli/deepagents_cli/shell.py) that exposed a shell tool with a per-command timeout: int | None = None parameter. The LLM could pass timeout=300 on any individual command to override the default 120s.

#1107 replaced ShellMiddleware with the SDK's LocalShellBackend, which unified shell execution under the execute tool and made it available to subagents. However, LocalShellBackend.execute() and SandboxBackendProtocol.execute() only accept command: str -- the per-command timeout parameter was list. The timeout is fixed at init time (self._timeout) and cannot be overridden per call ...so the LLM lost the ability to extend the timeout for long-running commands.

This PR restores that capability at the SDK level:

  • Add keyword-only timeout: int | None = None parameter to execute() across SandboxBackendProtocol and all implementations
  • Update timeout error message to guide LLM agents toward using the timeout parameter
  • Catch ValueError in middleware alongside NotImplementedError to prevent unhandled exceptions from crashing the agent loop

(negligible) Breaking changes

LocalShellBackend.__init__ timeout type narrowed from float to int.

🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.

## 📋 Pull Request Information **Original PR:** https://github.com/langchain-ai/deepagents/pull/1154 **Author:** [@mdrxy](https://github.com/mdrxy) **Created:** 2/6/2026 **Status:** 🔄 Open **Base:** `main` ← **Head:** `mdrxy/execute-timeout` --- ### 📝 Commits (3) - [`e98607b`](https://github.com/langchain-ai/deepagents/commit/e98607be605204ff5f7fc4976fb1dc9a7ab9cfd3) feat(sdk): add per-command timeout override to execute() - [`5b6cc15`](https://github.com/langchain-ai/deepagents/commit/5b6cc1576896acf82995377226ca2d62b4f6cf62) Merge remote-tracking branch 'origin/main' into mdrxy/execute-timeout - [`41c7d28`](https://github.com/langchain-ai/deepagents/commit/41c7d28f16920eaac165b33f71bc17b6f3602bbe) add 1hr max timeout ### 📊 Changes **21 files changed** (+521 additions, -55 deletions) <details> <summary>View changed files</summary> 📝 `libs/cli/deepagents_cli/backends.py` (+15 -3) 📝 `libs/cli/deepagents_cli/integrations/daytona.py` (+10 -1) 📝 `libs/cli/deepagents_cli/integrations/langsmith.py` (+6 -2) 📝 `libs/cli/deepagents_cli/integrations/modal.py` (+10 -1) 📝 `libs/cli/deepagents_cli/integrations/runloop.py` (+7 -1) 📝 `libs/cli/tests/unit_tests/test_backend_timeout.py` (+1 -1) 📝 `libs/deepagents/deepagents/backends/__init__.py` (+2 -1) 📝 `libs/deepagents/deepagents/backends/composite.py` (+21 -12) 📝 `libs/deepagents/deepagents/backends/local_shell.py` (+44 -7) 📝 `libs/deepagents/deepagents/backends/protocol.py` (+13 -1) 📝 `libs/deepagents/deepagents/backends/sandbox.py` (+5 -0) 📝 `libs/deepagents/deepagents/middleware/filesystem.py` (+39 -2) 📝 `libs/deepagents/tests/integration_tests/test_filesystem_middleware.py` (+3 -3) 📝 `libs/deepagents/tests/unit_tests/backends/test_composite_backend_async.py` (+2 -2) 📝 `libs/deepagents/tests/unit_tests/backends/test_sandbox_backend.py` (+1 -1) 📝 `libs/deepagents/tests/unit_tests/test_local_sandbox_operations.py` (+7 -3) ➕ `libs/deepagents/tests/unit_tests/test_local_shell.py` (+102 -0) 📝 `libs/deepagents/tests/unit_tests/test_middleware.py` (+199 -4) 📝 `libs/deepagents/tests/unit_tests/test_middleware_async.py` (+6 -6) 📝 `libs/harbor/deepagents_harbor/backend.py` (+13 -2) _...and 1 more files_ </details> ### 📄 Description Prior to #1107, the CLI used a standalone `ShellMiddleware` (in `libs/cli/deepagents_cli/shell.py`) that exposed a `shell` tool with a per-command `timeout: int | None = None` parameter. The LLM could pass `timeout=300` on any individual command to override the default 120s. #1107 replaced `ShellMiddleware` with the SDK's `LocalShellBackend`, which unified shell execution under the `execute` tool and made it available to subagents. However, `LocalShellBackend.execute()` and `SandboxBackendProtocol.execute()` only accept `command: str` -- the per-command timeout parameter was list. The timeout is fixed at init time (`self._timeout`) and cannot be overridden per call ...so the LLM lost the ability to extend the timeout for long-running commands. This PR restores that capability at the SDK level: - Add keyword-only `timeout: int | None = None` parameter to `execute()` across `SandboxBackendProtocol` and all implementations - Update timeout error message to guide LLM agents toward using the timeout parameter - Catch `ValueError` in middleware alongside `NotImplementedError` to prevent unhandled exceptions from crashing the agent loop ## (negligible) Breaking changes `LocalShellBackend.__init__` timeout type narrowed from `float` to `int`. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>
yindo added the pull-request label 2026-02-16 09:18:20 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
MCP
backends
bug
cli
config
customization
deepagents
documentation
external
feature
fix
good first issue
help wanted
integration
internal
invalid
middleware
model
needs investigation
p0
p1
p2
p3
performance
pull-request
Mirrored from GitHub Pull Request
 question
sdk
subagents
unable to reproduce
windows
No Label pull-request
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/deepagents#1175
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?