Experimental Code Coverage Orchestrator
Testing & Qualityexperimental-code-coverage-orchestratorExperimental Code Coverage Orchestrator is an Agent Skill for testing & quality published in the chromium/chromium repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.
Triages Chrome code coverage bugs, initializes the environment, and delegates to the appropriate debugging sub-agent (CQ or CI) using a shared state file.
QUICK START
How to use this skill
Bring this guide into your coding agent with a prompt tailored to the tool you use.
- Open your project in Codex.
- Copy and paste the prompt below.
- Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/chromium/chromium/blob/HEAD/agents/skills/experimental-code-coverage-orchestrator/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/experimental-code-coverage-orchestrator, but wait for my approval before writing files or running scripts.
Copying this prompt does not install or run the skill. Review third-party files before use.
Chrome Code Coverage Triage Orchestrator
You are the entry point for triaging Chrome code coverage bugs. Your job is to:
- Gather initial information from the bug.
- Set up the debugging environment.
- Determine if the issue is for the Commit Queue (CQ) or Continuous Integration (CI).
- Invoke the correct specialized sub-agent to handle the detailed debugging, using a state file for communication.
- Report the final outcome based on the sub-agent's results in the state file.
[!IMPORTANT] Sequential Execution & Status Updates:
- Sequential Execution: Execute steps in order. Do not proceed until the current step is fully complete.
- State File: All key information, inputs, and outputs for sub-agents must be written to and read from
scratch/triage_state.json.- Status Updates: You MUST update the
statusfield inscratch/triage_state.jsonwhenever transferring to another step (i.e., at the end of each step) to track progress. The status value should reflect the step name or number just completed (e.g.,"Orchestrator: Step 2 Completed"). The final terminal status values must be set to either"FIXED"or"ESCALATED".
Workflow Steps
1. Ingest Bug & Extract CL
- Action: Get the Buganizer ID from the user.
- Tool: Use the
buganizerrender_issue_with_externaltool to fetch bug details (title, description, and all comments) to avoid external issue redaction. - Extract: Find Gerrit CL URLs
(
chromium-review.googlesource.com/c/chromium/src/+/...). - Gate: If NO Gerrit CL URL is found, STOP and ask the user to provide one.
- Initialize State: Create
scratch/triage_state.jsoninitialized from the template defined in templates/triage_state_template.json. Store the extracted Buganizer ID inbug_id, the Gerrit CL URL inoriginal_cl, and the retrievedtitle,description, andcommentslist underbug_details.
2. Assess Issue Scope
- Action: Based on
bug_detailsinscratch/triage_state.json, determine the issue type:- CQ Issue: Mentions Gerrit UI, per-CL coverage. Update
issue_typeto "CQ". - CI Issue: Mentions
go/cr-coverage, dashboards. Updateissue_typeto "CI". - Out of Scope: If neither, inform the user and STOP.
- CQ Issue: Mentions Gerrit UI, per-CL coverage. Update
3. Initialize Environment
- Action: Ensure the workspace is ready and coverage tools are present.
- Skill: Use the
experimental-code-coverage-installerskill (SKILL.md) to verify and install dependencies.
4. Delegate to Sub-Agent
- Gate: Verify that the environment has been successfully initialized
(i.e.,
statusinscratch/triage_state.jsonis"Orchestrator: Step 3 Completed"). Do NOT proceed to invoke the sub-agent if the environment initialization failed. - Action: Invoke the appropriate sub-agent based on
issue_typefrom the state file.- If CQ Issue: Launch a
cq_debuggersub-agent to investigate the failure. Equip the sub-agent with theexperimental-code-coverage-cq-debuggerskill to execute the CQ coverage triaging playbook againstscratch/triage_state.json. - If CI Issue: Notify user CI triage is not yet supported and STOP.
- If CQ Issue: Launch a
5. Report Final Outcome
- Action: Once the sub-agent completes, read
scratch/triage_state.jsonto verify completion. - Provide Final Summary Report: Return
scratch/final_hypothesis_summary.mddirectly to the user by presenting a clickable file link toscratch/final_hypothesis_summary.md(which contains the final hypothesis summary, Root Cause Analysis, debugging steps taken, and full Buildbucket tryjob URLs across iterations). Do NOT re-summarize or synthesize the findings. - Clean up: Prompt the user if they would like to remove
scratch/triage_state.jsonafter reporting.