AgentStack
SKILL verified MIT Self-run

Update Claude Md

skill-bongrealty-skillcraft-update-claude-md · by bongrealty

After any correction from the user, update CLAUDE.md with a rule to prevent the same mistake from recurring. Use when the user corrects your behavior or says "don't do that again". Ruthlessly iterate CLAUDE.md until mistake rate measurably drops.

No reviews yet
0 installs
11 views
0.0% view→install

Install

$ agentstack add skill-bongrealty-skillcraft-update-claude-md

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Update Claude Md? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

After every correction the user gives you, do the following:

Steps

  1. Identify the mistake pattern — What went wrong? What should have been done instead? Be specific.
  1. Write a rule — Formulate it as a clear, actionable rule that prevents this class of mistake. Lead with the rule, then add context if needed.
  1. Update CLAUDE.md — Append the rule to the relevant CLAUDE.md (project-level if in a project, global ~/.claude/CLAUDE.md if it's a general behavior). Add it under a "Corrections" or "Rules" section.
  1. Confirm — Tell the user: "Updated CLAUDE.md with: [rule summary]"

Principles

  • After EVERY correction, end the session with: "Update your CLAUDE.md so you don't make that mistake again."
  • Ruthlessly edit CLAUDE.md over time — if a rule isn't reducing mistakes, rewrite it
  • If the project has a /memory or tasks/ notes directory, update it there too and point CLAUDE.md at it
  • Claude is eerily good at writing rules for itself — trust the process

Rule Format

## Rules / Corrections

- [Rule]: [What to do / not do]. [Why: brief context from the mistake]

Example

User: "Stop wrapping every response with a summary of what you just did."

Claude updates CLAUDE.md:

- No trailing summaries: Do not summarize actions at the end of responses. The user can read the diff.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.