[t3dotgg]

Problem

In the MCP tool-use eval (#24635), GPT-6.1 Sol polled delegates that were still running. The old running-task next text let it wait when it "needed the result now". Sol also left a stale review delegate running after it started a new one.

Change

Three one-line text changes:

• packages/contracts/src/orchestratorMcp.ts: the running-task next now says "Completion wakes this thread, so end your turn. Do not wait on it or poll it."
• apps/server/src/mcp/toolkits/orchestrator/tools.ts: the task_status description says not to call it on a running task before its completion wakes you.
• packages/provider-core/src/server/orchestrationInstructions.ts: stop a delegate run you replace or no longer need with task_cancel.

Scope and approval

Maintainer work (Theo). Text only, no behavior change. It follows the instruction cuts in #24850.

Verification

Full MCP eval runs with this text (a case fails when half or more of its samples fail):

• Claude Opus 5.5: 1 failing case of 85, 95% of samples pass. The failing case is child-pr-not-linked: the child runs out of its 8 steps.
• GPT-6.1 Sol: 0 failing cases of 83. The 3 Claude-only cases do not run.
• GPT-6 Astra, on the 7 delegate cases this text touches: 14 of 14 samples pass.

node scripts/check-changed.ts passes. vp test run src/orchestration-v2/testkit src/mcp in apps/server passes (413 tests).

Changes made with Claude Opus 5.5 in Claude Code (T3 Code).

🤖 Generated with Claude Code

[macroscopeapp]

Approvability

Verdict: Needs Human Approval

Macroscope's review found this PR requires human approval — This small change updates the agent-facing running-delegate protocol and the default orchestration instructions used for MCP-enabled agents. Because it changes default product behavior, human review is required under the configured approval criteria.

^{You can add or adjust custom eligibility rules. Learn more.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	5.0 KiB	5.0 KiB	−7 B (−0.1%)	6.8 KiB	✅
Codex	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Codex	Live turn WebSocket wire	1.2 KiB	1.2 KiB	−7 B (−0.6%)	2.0 KiB	✅
Codex	Live turn WebSocket decoded	20.9 KiB	20.9 KiB	0 B (0.0%)	29.3 KiB	✅
Codex	Live turn messages	2	2	0 (0.0%)	8	✅
Claude	Total thread wire	5.0 KiB	5.0 KiB	−24 B (−0.5%)	6.8 KiB	✅
Claude	Thread snapshot wire	3.8 KiB	3.8 KiB	0 B (0.0%)	4.9 KiB	✅
Claude	Live turn WebSocket wire	1.2 KiB	1.2 KiB	−24 B (−2.0%)	2.0 KiB	✅
Claude	Live turn WebSocket decoded	21.2 KiB	21.2 KiB	−41 B (−0.2%)	29.3 KiB	✅
Claude	Live turn messages	2	1	−1 (−50.0%)	8	✅

Baseline: 3d00499 · PR result: 6c9ea4d · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 108.5 KiB
• Claude decoded thread snapshot: 108.8 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[coderabbitai]

No actionable comments were generated in the recent review.

Recent review info

Run configuration

• Source mode: Auto
• Repository: pingdotgg/t3code
• Branch: copy/delegate-wake-copy
• File: .coderabbit.config.ts
• Commit: 6c9ea4d
• Layers: Repository TS config
• Review profile: CHILL
• Plan: Team
• Run ID: ced4a3e4-6b13-42de-b267-b219bba07196

Commits

Reviewing files that changed from the base of the PR and between 3d00499 and 6c9ea4d.

Files selected for processing: 3

• apps/server/src/mcp/toolkits/orchestrator/tools.ts
• packages/contracts/src/orchestratorMcp.ts
• packages/provider-core/src/server/orchestrationInstructions.ts

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 8 remain after this review.

Walkthrough

The changes update orchestration guidance for running-task polling and delegate cancellation. Callers are instructed to end their turn when completion will wake the thread, and to cancel delegate runs only when replacing them or when they are no longer needed.

Changes: Orchestration guidance

Layer / File(s)	Summary
Running-task polling guidance packages/contracts/src/orchestratorMcp.ts, apps/server/src/mcp/toolkits/orchestrator/tools.ts	The running-task message says completion wakes the thread and directs callers not to wait or poll. The task-status description says not to query a running task before completion wakes the caller.
Delegate cancellation guidance packages/provider-core/src/server/orchestrationInstructions.ts	The instruction says to use task_cancel when replacing a delegate run or when it is no longer needed. The prohibition on using t3_thread_interrupt for a childThreadId remains.

Priority: Low

Estimated code review effort: 2 (Simple) | ~8 minutes

Change: Bug fix

Possibly related PRs

• pingdotgg/t3code#24782: It adds the running-task message that this PR updates, alongside matching task-status guidance.
• pingdotgg/t3code#24850: It introduced the orchestration guidance that this PR refines.
• pingdotgg/t3code#21676: It changes cancellation finalization for delegated runs, related to this PR’s delegate-cancellation instruction.

Suggested labels: bugfix

Suggested reviewers: maria-rcks, macroscopeapp

Merge Risk: Minimal · up to 6c9ea

No actionable merge-blocking risk was identified.

Pre-merge checks: 4

Check name	Status	Explanation
Title check		The title clearly identifies the main change: running delegates now tell agents to end the turn. It is concise and related to the instruction updates.
Description check		The description includes the required Problem, Change, Scope and approval, and Verification sections. It explains the affected instructions, approval context, focused evaluation results, and test resu…
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Full details: Description check

Explanation

The description includes the required Problem, Change, Scope and approval, and Verification sections. It explains the affected instructions, approval context, focused evaluation results, and test results.

Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}