Skip to content

chore: promote staging to staging-promote/b952d229-23331469361 (2026-03-20 07:18 UTC)#1459

Open
ironclaw-ci[bot] wants to merge 2 commits intostaging-promote/b952d229-23331469361from
staging-promote/c1762616-23332963145
Open

chore: promote staging to staging-promote/b952d229-23331469361 (2026-03-20 07:18 UTC)#1459
ironclaw-ci[bot] wants to merge 2 commits intostaging-promote/b952d229-23331469361from
staging-promote/c1762616-23332963145

Conversation

@ironclaw-ci
Copy link
Contributor

@ironclaw-ci ironclaw-ci bot commented Mar 20, 2026

Auto-promotion from staging CI

Batch range: c4ab382522c86e7e19d55fee760b125fb1970518..c17626160ce956a5e7c64a59b3e65c1801fee21f
Promotion branch: staging-promote/c1762616-23332963145
Base: staging-promote/b952d229-23331469361
Triggered by: Staging CI batch at 2026-03-20 07:18 UTC

Commits in this batch (12):

Current commits in this promotion (2)

Current base: staging-promote/b952d229-23331469361
Current head: staging-promote/c1762616-23332963145
Current range: origin/staging-promote/b952d229-23331469361..origin/staging-promote/c1762616-23332963145

Auto-updated by staging promotion metadata workflow

Waiting for gates:

  • Tests: pending
  • E2E: pending
  • Claude Code review: pending (will post comments on this PR)

Auto-created by staging-ci workflow

vnz and others added 2 commits March 19, 2026 23:22
#1426)

* fix: register sandbox jobs in ContextManager for query tool visibility

Sandbox jobs created via execute_sandbox() were persisted to the database
but never registered in the in-memory ContextManager. Since all query tools
(list_jobs, job_status, job_events, cancel_job) only search the
ContextManager, sandbox jobs were invisible to the agent despite running
successfully in Docker containers.

Changes:
- Add register_sandbox_job() to ContextManager (pre-determined UUID,
  starts InProgress, respects max_jobs)
- Extract insert_context() helper to deduplicate create_job_for_user
  and register_sandbox_job
- Add update_context_state / update_context_state_async to sync
  ContextManager state on sandbox job completion/failure
- Extend job_monitor with spawn_job_monitor_with_context() and
  spawn_completion_watcher() so fire-and-forget jobs transition out
  of InProgress when the container finishes
- Make CancelJobTool sandbox-aware (stops container + updates DB)
- Wire sandbox deps into CancelJobTool in register_job_tools()
- 8 regression tests across context manager and job monitor

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: add missing allow_always field in PendingApproval test literal

Upstream commit 09e1c97 added the allow_always field to PendingApproval
but missed updating the test struct literal, breaking compilation.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Bedrock uses IAM credentials (instance roles, env vars, SSO) resolved
by the AWS SDK at call time, so `provider` is never set during startup.
Exclude it from the post-init validation that checks for missing API keys.

Closes #1009

Co-authored-by: brajul <brajul@amazon.com>
Co-authored-by: Illia Polosukhin <ilblackdragon@gmail.com>
@github-actions github-actions bot added scope: agent Agent core (agent loop, router, scheduler) scope: tool Tool infrastructure scope: tool/builtin Built-in tools size: XL 500+ changed lines risk: medium Business logic, config, or moderate-risk modules contributor: core 20+ merged PRs labels Mar 20, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

contributor: core 20+ merged PRs risk: medium Business logic, config, or moderate-risk modules scope: agent Agent core (agent loop, router, scheduler) scope: tool/builtin Built-in tools scope: tool Tool infrastructure size: XL 500+ changed lines staging-promotion

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants