agentleFS
mhihasanmhihasan/agentic-sdlc/skills/receiving-plan-review/SKILL.md

receiving-plan-review

Skill1 starsChanged 5 months agoMITView on GitHub

Use when a reviewing-plan verdict is DO NOT PROCEED or PROCEED WITH CHANGES and findings need to be worked through — before editing the plan. Triggers on "address plan findings", "fix the plan review", "work through plan blockers", or when reviewing-plan emits findings. Verify each finding against the ticket and codebase before accepting it.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Scope gap (missing AC): Read the ticket. Does that AC actually exist verbatim?

    line 30
  2. Distinctive:

    Scope creep (extra work): Check the ticket ACs. If the plan work is traceable to any AC (even indirectly), push back with the AC reference.

    line 32
  3. Distinctive:

    When a finding holds up, make a targeted edit to the plan:

    line 54
  4. Very specific:

    After each edit, note what changed: "Fixed: added T6 for [AC description]."

    line 60
  5. Very specific:

    Do not re-run reviewing-plan yourself — the developer triggers the next review.

    line 68

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

receiving-plan-review

Work through reviewing-plan findings with technical rigor.

Core principle: A finding is a claim, not an order. Verify it against the ticket and codebase before touching the plan. Verify each finding before fixing it. Push back when findings are wrong.

The Pattern

FOR each finding (BLOCKER first, then SHOULD-FIX, then NIT):
  1. READ the finding completely
  2. VERIFY against the ticket AC and codebase
  3. EVALUATE: does the finding hold up?
  4. DECIDE: accept, push back, or ask
  5. ACT: edit the plan if accepted; state reasoning if rejected

NEVER accept a finding without verifying it first. The judge ran with limited context. You have the full codebase.

Verification by Finding Type

Scope gap (missing AC): Read the ticket. Does that AC actually exist verbatim? If yes — add the task. If the finding misread the ticket, push back with the exact ticket text.

Scope creep (extra work): Check the ticket ACs. If the plan work is traceable to any AC (even indirectly), push back with the AC reference. If genuinely untraceable — accept.

Codebase grounding (file/export doesn't exist): Grep the repo. If it doesn't exist — accept. If it does — push back with the path and line number.

Over-engineering: Is the abstraction actually introduced? Does the codebase already use this pattern everywhere? If yes, push back — consistency is a valid reason. If genuinely disproportionate — accept.

Breaking change: Trace the consumers yourself. If the finding identified real unaddressed consumers — accept. If the change is additive or the consumers are already handled — push back with the evidence.

Task decomposition: Would a reviewer meaningfully approve one task while rejecting its neighbor? If tasks are genuinely entangled — accept. If the finding is splitting for its own sake — push back.

Pushing Back

Push back when:

  • The finding misread the ticket (cite the exact AC text)
  • The finding flagged a file/export that does exist (cite path + line)
  • The finding flagged scope creep that is ticket-traceable (cite the AC)
  • The proposed fix introduces more complexity than the original problem

How: one sentence, evidence first. No over-explaining.

Accepting a Finding

When a finding holds up, make a targeted edit to the plan:

  • Scope gap → add the missing task
  • Scope creep / over-engineering → remove or simplify the task
  • Grounding error → fix the file path or method name
  • Silent breaking change → add a migration/compat step

After each edit, note what changed: "Fixed: added T6 for [AC description]."

After All Findings Are Addressed

Once all BLOCKERs and SHOULD-FIXes are resolved or pushed back with reasoning:

"All findings addressed. Re-run /reviewing-plan to get a fresh verdict."

Do not re-run reviewing-plan yourself — the developer triggers the next review.

Common Mistakes

MistakeFix
Accepting all findings without reading the ticketRead the AC first. The judge can be wrong.
Editing the plan before verifyingVerify first, always.
Pushing back without evidenceCite the ticket text, file path, or codebase pattern.
Fixing NITs before BLOCKERsBLOCKERs first. NITs are the developer's call.
Re-running reviewing-plan yourselfThe developer triggers the next review.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.