anomalyco/opencode
1
0
Fork 0
mirror of https://github.com/anomalyco/opencode.git synced 2026-07-23 02:36:57 -04:00
Code Issues 5.9k Packages Projects Releases 19 Wiki Activity

[PR #12136] fix(acp): handle permission requests from child sessions #14082

New Issue
Open
opened 2026-02-16 18:18:53 -05:00 by yindo · 0 comments
yindo commented 2026-02-16 18:18:53 -05:00
Owner
Copy Link
Copy Source

Original Pull Request: https://github.com/anomalyco/opencode/pull/12136

State: open
Merged: No

When Task subagents create child sessions, they are not registered in the ACP session manager. This caused permission requests from those sessions to be silently ignored, freezing the session indefinitely.

What does this PR do?

Fixes #12133. Now fetches session info from SDK when the session is not found in the manager, allowing permission requests from child sessions to be properly forwarded to the ACP client.

How did you verify your code works?

I created a super simple ACP test client in Python that tests permission requests from Task subagents are properly forwarded to the ACP client (which auto-accepts the permission) by spawning 4 parallel OpenCode ACP processes (2 prod, 2 dev) with two prompts:
Baseline:

What is 2+2?

Test:

Use the Task tool to create a subagent that writes a test file to /tmp/test.txt with the content
"hello world", then reads it back and reports the contents.

======================================================================
SUMMARY
======================================================================
Name       Command                   Status     Perms  Child  Time
----------------------------------------------------------------------
prod-simple opencode acp              OK         0      0      3.5s
prod-perm  opencode acp              FAIL: Timeout 0      0      120.9s
dev-simple bun run dev acp           OK         0      0      7.3s
dev-perm   bun run dev acp           OK         2      2      20.4s
----------------------------------------------------------------------
Passed: 3/4
With child session permissions: 1/4

The prod-perm is EXPECTED to fail and time out (because the permissions never get forwarded, thus the ACP client just doesn't ever know it needs to do something)

test_acp.py

**Original Pull Request:** https://github.com/anomalyco/opencode/pull/12136 **State:** open **Merged:** No --- When Task subagents create child sessions, they are not registered in the ACP session manager. This caused permission requests from those sessions to be silently ignored, freezing the session indefinitely. ### What does this PR do? Fixes #12133. Now fetches session info from SDK when the session is not found in the manager, allowing permission requests from child sessions to be properly forwarded to the ACP client. ### How did you verify your code works? I created a super simple ACP test client in Python that tests permission requests from Task subagents are properly forwarded to the ACP client (which auto-accepts the permission) by spawning 4 parallel OpenCode ACP processes (2 prod, 2 dev) with two prompts: **Baseline:** ``` What is 2+2? ``` **Test:** ``` Use the Task tool to create a subagent that writes a test file to /tmp/test.txt with the content "hello world", then reads it back and reports the contents. ``` ``` ====================================================================== SUMMARY ====================================================================== Name Command Status Perms Child Time ---------------------------------------------------------------------- prod-simple opencode acp OK 0 0 3.5s prod-perm opencode acp FAIL: Timeout 0 0 120.9s dev-simple bun run dev acp OK 0 0 7.3s dev-perm bun run dev acp OK 2 2 20.4s ---------------------------------------------------------------------- Passed: 3/4 With child session permissions: 1/4 ``` The `prod-perm` is EXPECTED to fail and time out (because the permissions never get forwarded, thus the ACP client just doesn't ever know it needs to do something) [test_acp.py](https://github.com/user-attachments/files/25074621/test_acp.py)
yindo added the pull-request label 2026-02-16 18:18:53 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
acp
bad
bug
discussion
docs
documentation
enhancement
extension
good first issue
growth
help-wanted
mcp
model-problem
needs:compliance
nix
opentui
perf
pull-request
Mirrored from GitHub Pull Request
 tui
web
windows
zen
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: anomalyco/opencode#14082
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?