Skip to content

Fail incomplete agent sandbox runs#549

Merged
chubes4 merged 1 commit into
mainfrom
fix/issue-534-agent-sandbox-incomplete
Jun 3, 2026
Merged

Fail incomplete agent sandbox runs#549
chubes4 merged 1 commit into
mainfrom
fix/issue-534-agent-sandbox-incomplete

Conversation

@chubes4
Copy link
Copy Markdown
Collaborator

@chubes4 chubes4 commented Jun 3, 2026

Summary

  • Treat nested agent sandbox results as runtime failures when the nested agents/chat payload is still processing, has pending tools, or exhausts max turns without a terminal answer.
  • Preserve the existing agent_runtime_incomplete failure contract while expanding detection beyond agent_runtime.result to raw nested payloads captured in stdout/stderr.
  • Add smoke coverage for the Homeboy-style pending-tools payload and max-turn exhaustion payloads.

Trackers

Verification

  • npm run build
  • npm run agent-runtime-failure-smoke
  • npm run wordpress-plugin-smoke

Known local verification blocker

  • npm run check could not complete reliably in this local Studio/Kimaki environment because artifact-contract-smoke repeatedly collided with long-lived/orphaned local Playground/OpenCode listener processes on random localhost ports. The focused checks above pass on the rebased branch.

AI assistance

  • AI assistance: Yes
  • Tool(s): OpenCode (openai/gpt-5.5)
  • Used for: Drafted the code changes and smoke coverage; Chris remains responsible for review and final validation.

@chubes4 chubes4 merged commit 36082e8 into main Jun 3, 2026
@chubes4 chubes4 deleted the fix/issue-534-agent-sandbox-incomplete branch June 4, 2026 20:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Agent sandbox run should fail or timeout when nested agent remains processing with pending tools

1 participant