docs(#294): add cross-repo ordering guidance to PR body instructions

[fullsend-ai-coder]

Summary

• Add "Cross-repo ordering" guidance to step 10d of skills/code-implementation/SKILL.md
• When a motivating issue references a different repository, the PR summary should lead with what files in the current repo are changing and why
• Cross-repo references (upstream issues, motivation) should appear in a secondary "Context" or "Related" subsection, not the opening sentence

Motivation

PR #273 was closed by a reviewer who appeared confused about which repository the PR targeted, likely because the summary's first sentence referenced an external repo (fullsend-ai/fullsend#5273) rather than describing the local file changes. The existing step 10d guidance covers what to include in a PR body but not how to order local changes vs. cross-repo context.

Testing

• This is a documentation-only change to skill guidance prose
• Validated with gitlint (passed), secret scan (passed)
• Pre-commit could not run in sandbox (network blocked); the post-script runs it authoritatively

---

Closes #294

Post-script verification

• Branch is not main/master (agent/294-pr-body-local-first)
• Secret scan passed (gitleaks — fc7b83e1e64b3d786b2d500189c142759fa06451..HEAD)
• PR body secret scan passed (gitleaks — no-git)
• Pre-commit hooks passed (authoritative run on runner)
• Tests ran inside sandbox

[rh-hemartin]

/fs-review

[fullsend-ai-review]

🤖 Finished Review · ✅ Success · Started 6:38 AM UTC · Completed 6:46 AM UTC
Commit: 7ada4e0 · View workflow run →

[fullsend-ai-review]

Review

Findings

Medium

• [protected-path] skills/code-implementation/SKILL.md — This file is under skills/, a protected path. The PR links to issue Code agent PR summary should lead with local file changes, not cross-repo references #294 and the description explains the rationale for adding cross-repo ordering guidance. Human approval is required for all protected-path changes regardless of review agent verdict.

---

Labels: PR is a documentation-only change created by the code agent

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 7:10 AM UTC · Completed 7:18 AM UTC
Commit: 04181b2 · View workflow run →

[fullsend-ai-retro]

Retro: PR #295 — Add cross-repo ordering guidance

Verdict: Clean workflow, no new proposals.

This PR exemplifies the retro→triage→code→review pipeline working as designed. The retro agent identified a real problem (PR #273 was closed by a confused reviewer because the PR summary led with a cross-repo reference), filed issue #294, and the code agent implemented a correct, minimal fix in a single iteration — 8 lines of documentation guidance added to skills/code-implementation/SKILL.md.

Timeline

Time	Event
Jul 20 06:24	Triage agent processes issue #294 (run 29721580940)
Jul 20 06:28	Code agent starts (run 29721783826)
Jul 20 06:34	PR #295 created — 8 lines added, 0 deleted, 1 file changed
Jul 22 06:37	Human (rh-hemartin) approves and triggers /fs-review
Jul 22 06:38	Review agent starts (run 29897353078)
Jul 22 06:47	Review agent completes — protected-path flag, no code issues
Jul 22 07:07	PR merged via merge queue

What went well

• First-time-right implementation. No fix cycles, no rework. The code agent produced a clean, minimal change that passed review without modification.
• Independent verification by both agents. Issue Code agent PR summary should lead with local file changes, not cross-repo references #294 mentioned checking fix-review/SKILL.md for analogous PR body instructions. Both the code agent and review agent independently investigated this and correctly determined it was not needed — fix-review uses structured JSON output, not free-form PR descriptions.
• Low cost. Code agent: $1.82 / 2m41s. Review agent: $1.65 / ~9min. Total pipeline cost under $4.
• Surgical scope. The change added exactly what was needed — no scope creep, no unnecessary refactoring.

Known friction: label mismatch delays review dispatch

The 2-day gap between PR creation (Jul 20) and review (Jul 22) is partly caused by a known label naming mismatch: the repo has ready-to-review but dispatch routing checks for ready-for-review. The post-code script fails silently when trying to apply the non-existent label, so the review agent is never auto-dispatched. This is already covered by existing issues:

• Rename ready-to-review label to ready-for-review to match dispatch routing #351 — Rename ready-to-review to ready-for-review to match dispatch routing (labeled ready-to-code)
• Post-code.sh should auto-create ready-for-review label before applying it #274 — Post-code.sh should auto-create ready-for-review label before applying it
• Post-code script should create the ready-for-review label if it does not exist in the target repo #199 — Post-code script should create the label if it doesn't exist in the target repo

This retro provides additional evidence: PR #295 is another instance where the label mismatch prevented auto-dispatch, requiring manual /fs-review and adding ~48 hours of latency. However, the 2-day gap is also consistent with this repo's normal batch-review pattern (median 2-3 days for bot-authored PRs).

No new proposals

The workflow executed cleanly. The only systemic friction point (label mismatch preventing review auto-dispatch) is already well-documented across three open issues with a fix queued. No new proposals are warranted.