get-feedback-markdown-plan

name	get-feedback-markdown-plan
version	1.0.0
description	DEPRECATED — use the pair-* skill series (e.g. pair-consult / pair-optimize) instead. Evaluates inline feedback annotations on a markdown plan by verifying each critique against the codebase and context using subagents.

Get Feedback on Markdown Plan

⚠️ Deprecated. Use the pair-* skill series (e.g. pair-consult, pair-optimize) instead. This skill is no longer installed and is kept only for reference.

You have a markdown plan with inline feedback annotations (e.g., [CODEX], [Claude], or similar tags). Your job is to verify whether each annotation is correct by checking it against the codebase and project context.

Workflow

1. Parse — Extract all inline annotations from the plan. For each one, capture the annotation text and the plan section it refers to.

2. Group — Cluster related annotations that share context (e.g., multiple comments about the same service, same migration step, or same data flow). Each group becomes one subagent task.

3. Verify — Launch one subagent per group in parallel. Each subagent receives:

   • The plan section(s) under review
   • The annotation(s) to verify
   • Instruction: explore the codebase to determine if the critique is factually correct. Return one of:
     • Correct — the critique identifies a real problem. Briefly explain why.
     • Wrong — the critique is incorrect or based on a false assumption. Explain what it got wrong.
     • Inconclusive — cannot verify from available context. State what's missing.

4. Report — Present a summary table:

   #	Annotation	Verdict	Reason
   1	...	Correct / Wrong / Inconclusive	One-line explanation

   Then list the annotations that were Wrong with full reasoning, so the user knows which feedback to ignore.