[CI Failure Doctor] CI Failure Investigation - Run #37215

[github-actions]

CI Failure Investigation - Run #37215

Summary

Wasm golden fixtures for the basic-copilot workflow are still locked to Copilot CLI 0.0.412, but the build-wasm job now installs CLI 0.0.414, so TestWasmGolden_CompileFixtures/basic-copilot fails when the generated YAML drifts.

Failure Details

• Run: 22256948582
• Commit: 61b1c3a
• Trigger: push
• Failed job: build-wasm (TestWasmGolden_CompileFixtures/basic-copilot)

Root Cause Analysis

The wasm golden comparison reads the generated basic-copilot.lock.yml. The pipeline now installs Copilot CLI 0.0.414, which emits agent_version: "0.0.414" and runs /opt/gh-aw/actions/install_copilot_cli.sh 0.0.414, but the golden fixture still expects agent_version: "0.0.412" and the older install script. That mismatch trips the equality check in pkg/workflow/wasm_golden_test.go:98.

Failed Jobs and Errors

• build-wasm — TestWasmGolden_CompileFixtures/basic-copilot fails with a diff in basic-copilot:

  --- Expected
  +++ Actual
  @@ -246,3 +246,3 @@
                   version: "",
  -              agent_version: "0.0.412",
  +              agent_version: "0.0.414",
  ...
                - name: Install GitHub Copilot CLI
  -        run: /opt/gh-aw/actions/install_copilot_cli.sh 0.0.412
  +        run: /opt/gh-aw/actions/install_copilot_cli.sh 0.0.414
  

Investigation Findings

• The golden fixture has not been regenerated since the previous 0.0.412 expectation documented in issue 17466.
• Bumping the Copilot CLI or install script version without recomputing basic-copilot.lock.yml keeps the golden locked to the old CLI, so the string comparison in the wasm golden test repeatedly fails.

Recommended Actions

• Regenerate the wasm golden fixtures driven by TestWasmGolden_CompileFixtures (e.g., basic-copilot.lock.yml) so the expected output reflects Copilot CLI 0.0.414 (agent_version line and install_copilot_cli version).
• Add a checklist item or automation in the CLI bump/release process to rerun the wasm golden test and commit updated fixtures whenever the Copilot CLI or install script version changes.

Prevention Strategies

Add documentation explaining the dependency between pkg/workflow/wasm_golden_test.go and the embedded Copilot CLI/install script version. Include an explicit reminder to regenerate the goldens and revalidate TestWasmGolden_CompileFixtures during version bumps.

AI Team Self-Improvement

Update instructions.md (or the relevant AI prompt) with: "After bumping the Copilot CLI/install script version, regenerate the wasm golden lock files, rerun TestWasmGolden_CompileFixtures, and commit the new fixtures before pushing the change to avoid recurrent mismatches."

Historical Context

• Issue 17466 documented the same failure pattern for run [aw] Daily Project Performance Summary Generator (Using MCP Scripts) failed #37203. That investigation also resolved the failure by regenerating the wasm golden fixtures to match Copilot CLI 0.0.414, so this is a recurrence of that drift.

🩺 Diagnosis provided by CI Failure Doctor

To install this workflow, run gh aw add githubnext/agentics/workflows/ci-doctor.md@ea350161ad5dcc9624cf510f134c6a9e39a6f94d. View source at https://github.com/githubnext/agentics/tree/ea350161ad5dcc9624cf510f134c6a9e39a6f94d/workflows/ci-doctor.md.

• expires on Feb 22, 2026, 12:46 PM UTC