Skip to content

Update SKILL.md step 2 to require Recommended on follow-up replies #11

Open

Description

@fullsend-ai-retro

What happened

PR #9 updated the grillme agent definition (grillme.md) and the grilling skill (SKILL.md) to require a **Recommended:** <answer and why> line on inline questions. The agent definition at lines 225-226 explicitly covers follow-up thread replies: "When the follow-up is itself a decision question, include **Recommended:** the same way as new comments." However, SKILL.md step 2 still reads: "Reply in-thread when an answer needs a follow-up" — with no mention of including a recommendation. The review agent flagged this as a low-severity internal consistency finding in run 32518019927, but the PR was merged without addressing it.

What could go better

The SKILL.md and grillme.md now diverge on follow-up reply behavior. While grillme.md is the primary runtime document and the risk is low, SKILL.md serves as the canonical behavior spec for the grilling skill. If another agent or a future refactor consumes the skill definition without the grillme.md agent definition, follow-up replies would lack recommendations. This is a high-confidence finding — the review agent identified the exact gap and suggested a specific fix. The only uncertainty is whether the author deliberately chose not to address it (considering it negligible) or simply missed it during a fast merge cycle.

Proposed change

In .fullsend/custom/skills/grilling/SKILL.md, update step 2 of the Turn model from:

  1. Resolve threads whose answers are satisfactory. Reply in-thread when an answer needs a follow-up. Leave unanswered threads alone.

To something like:

  1. Resolve threads whose answers are satisfactory. Reply in-thread when an answer needs a follow-up — when that follow-up is itself a decision question, include **Recommended:** <answer and why> the same way as new comments. Leave unanswered threads alone.

This aligns SKILL.md with the instruction already present in grillme.md under "Rules for thread replies."

Validation criteria

After the change, SKILL.md step 2 and grillme.md's "Rules for thread replies" section should both mention including **Recommended:** on follow-up replies that are decision questions. Verify by grepping both files for Recommended — both should have matching coverage for new comments and follow-up replies.


Generated by retro agent from #9

Activity

  1. fullsend-ai-triage commented on Aug 21, 2026

    @fullsend-ai-triage

    🤖 Finished Triage · ✅ Success · Started 7:49 PM UTC · Completed 7:52 PM UTC

    Commit: aa08686 · View workflow run →

  2. fullsend-ai-triage commented on Aug 21, 2026

    @fullsend-ai-triage

    Confirmed the divergence: grillme.md lines under "Rules for thread replies" include the **Recommended:** instruction for follow-up decision questions, while SKILL.md step 2 omits it.

    Proposed test case

    # After fix, both files should mention Recommended for follow-ups
    grep -n 'Recommended' .fullsend/custom/skills/grilling/SKILL.md
    grep -n 'Recommended' .fullsend/custom/agents/grillme.md
    # Verify step 2 includes the follow-up clause
    grep -A2 'Reply in-thread' .fullsend/custom/skills/grilling/SKILL.md

    Single-line text change in SKILL.md step 2. No code or workflow changes needed.


    Next steps:

    • /fs-code — agent creates a PR to implement this issue
    • /fs-code <your instruction> — agent implements with your specific guidance
  3. fullsend-ai-coder commented on Aug 21, 2026

    @fullsend-ai-coder
    Contributor

    🤖 Finished Code · ✅ Success · Started 7:53 PM UTC · Completed 7:57 PM UTC

    Commit: aa08686 · View workflow run →

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions