skills.rest
patrickserrano
patrickserranoCOMMUNITY
Explore all skills by this author →
Website𝕏@patrickgserranoGitHub

claudemd-review

Reviews the current conversation for corrections and proposes a bounded CLAUDE.md edit.

3|Updated Jun 15, 2026
Search Tags:#claude code#review#conversation analysis#ai workflow#claude.md#project rules#diff proposal
One-click install
npx skills add https://github.com/patrickserrano/lacquer --skill claudemd-review-patrickserrano
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: claudemd-review
Source: https://github.com/patrickserrano/lacquer/tree/main/core/skills/claudemd-review
Command: npx skills add https://github.com/patrickserrano/lacquer --skill claudemd-review-patrickserrano

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a human corrects the AI on something a project's CLAUDE.md should already have covered, that lesson is often lost. This Skill captures the correction immediately and turns it into a concrete, minimal CLAUDE.md improvement proposal. ## Core Features & Use Cases - Conversation Scanning: Reviews the current session transcript for corrections, "no, actually" moments, and repeated clarifications. - Correction Classification: Sorts each friction point into already-covered-but-ignored, genuinely missing, or one-off cases so only durable conventions become edits. - Minimal Diff Proposals: Drafts a small, voice-matching CLAUDE.md diff and presents it for human review without applying anything automatically. - Use Case: After a long session where the user twice corrected the AI's commit message format, run this Skill to get a proposed CLAUDE.md line stating the exact convention, quoted against the corrections that motivated it. ## Quick Start Ask the AI to review this conversation for friction and propose a CLAUDE.md edit based on any corrections you made.

Frequently Asked Questions about claudemd-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn AI corrections into CLAUDE.md rules?▼

Run a CLAUDE.md review after a session with corrections. The Skill scans the transcript for friction, classifies each correction, and proposes a minimal diff stating the convention concretely, which you then review and apply yourself.

When should I review CLAUDE.md after a coding session?▼

Review right after being corrected on something CLAUDE.md should have covered, when explicitly asked to review it, or periodically after sessions with lots of back-and-forth. These moments signal rules that are missing or unclear.

claudemd-review vs skill-tuning-loop: which should I use?▼

Use claudemd-review for a quick single-session check when the ask is "check right now." Use skill-tuning-loop when friction repeats across many sessions and you want a validated, evidence-gated proposal tested against held-out cases.

Does the CLAUDE.md review apply edits automatically?▼

No. The Skill only presents a proposed diff with the motivating correction quoted. The human edits the file, because CLAUDE.md is read every session and a bad edit propagates everywhere.

What are the limitations of single-session CLAUDE.md review?▼

It only sees one conversation, so it can mistake a one-off correction for a durable pattern. One-off, task-specific friction is classified and excluded, but recurring patterns are better validated with skill-tuning-loop.

Related Skills

Explore More Software Engineering Skills→
GuitarAlchemistGuitarAlchemist

correct

Convert user corrections into permanent CLAUDE.md rules during AI sessions.

Official
Advanced
botminterbotminter
18

workspace-sync

Sync profile templates, team repositories, and member workspaces with diagnostics.

Official
Advanced
RohithdgrrRohithdgrr

skillopt-sleep

Consolidates past Claude sessions into gated CLAUDE.md and SKILL.md improvements.

Community
Advanced
m2ai-portfoliom2ai-portfolio
4

claudemd-router

Restructure bloated CLAUDE.md into a lean router delegating to .claude/rules files.

Official
Advanced