fix(test): add mock yq to prerequisites test setup (#47)

[fullsend-ai-coder]

The prerequisites-creates-allowed-issue test case fails when yq is not installed because post-triage.sh disables cross-repo issue creation when it cannot parse config.yaml. Add a mock yq binary to MOCK_BIN that reads the YAML fixture and returns allowlist values, following the existing pattern for mock gh and fullsend binaries. This makes the test hermetic — it passes regardless of whether yq is installed on the host.

---

Closes #47

Post-script verification

• Branch is not main/master (agent/47-mock-yq-prerequisite-test)
• Secret scan passed (gitleaks — 2540ede96537b32447321564bec60226efca120e..HEAD)
• Pre-commit hooks passed (authoritative run on runner)
• Tests ran inside sandbox

[github-actions]

Functional tests did not run

Functional tests run automatically for org/repo members and collaborators on pull requests.

For other contributors, a maintainer must add the ok-to-test label after the latest push.

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 9:20 PM UTC · Completed 9:30 PM UTC
Commit: 2f09703 · View workflow run →

[fullsend-ai-retro]

Retro: PR #50 — mock yq approach rejected in favor of simpler env fix

Timeline

1. 2026-07-07 22:42 UTC — Retro bot creates issue #47 from PR fix(triage): enforce question action and label consistency #40 analysis, identifying prerequisites-creates-allowed-issue test failure when yq is missing. Recommends mock yq binary (preferred) or skip test.
2. 2026-07-07 22:46 UTC — Triage agent completes, echoes mock approach, adds ready-to-code.
3. 2026-07-07 22:57 UTC — Code agent creates PR #50 with 46-line bash mock yq parser (~10 min implementation).
4. No review agent ran — PR had zero automated reviews over 22 hours.
5. 2026-07-08 21:19 UTC — Human (ralphbean) closes PR without merge, commenting: prefers simply ensuring yq is present in the test environment rather than mocking it.

Analysis

The retro agent's proposal was the root cause of the rejection. It framed the problem as "test lacks a mock dependency" and recommended a code-level mock, when the simplest fix was an environment change (install yq). The code agent faithfully implemented the retro's preferred approach, producing a 46-line bash YAML parser. The human wanted a ~1-line change. This is a different failure mode from approach preference order (covered by fullsend-ai/fullsend#1471) — here the issue body didn't include the simpler option at all.

Evidence for existing issues

• fullsend-ai/fullsend#3660: PR fix(test): add mock yq to prerequisites test setup (#47) #50 confirms the coder-bot PR review coverage gap — zero review agent runs on a bot-authored PR open for 22 hours.
• fullsend-ai/fullsend#3050: Issue fix(test): guard prerequisites test for missing yq dependency #47 retains ready-to-code after PR fix(test): add mock yq to prerequisites test setup (#47) #50 was closed by maintainer, making it eligible for re-dispatch.
• fullsend-ai/fullsend#2840: When the code agent is re-dispatched on issue fix(test): guard prerequisites test for missing yq dependency #47, it won't know PR fix(test): add mock yq to prerequisites test setup (#47) #50 was rejected or why — the human's feedback is only on the closed PR, not on the issue.

Proposals filed

• Retro proposals for test failures should consider environment-level fixes alongside code-level mocks (in fullsend-ai/agents)