Skip to content
Back to skills

Update Claude Md

ASecurity

Update CLAUDE.md based on what was just discussed in conversation

  • 23 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added May 28, 2026
toolsgo

Security analysis

A100/100

Scanned May 28, 2026

npx -y skills add tta-lab/ttal-cli --skill update-claude-md --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Update Claude Md?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Update Claude Md
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/tta-lab-update-claude-md/badge)](https://www.skillsdirectory.com/skills/tta-lab-update-claude-md)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: update-claude-md
description: "Update CLAUDE.md based on what was just discussed in conversation"
category: command
claude-code: {}
---

# Update CLAUDE.md

## Overview

After a meaningful conversation — decisions made, patterns discovered, preferences stated — the user may ask you to capture what was discussed into the project's CLAUDE.md. This ensures future sessions benefit from today's context.

## Which CLAUDE.md?

Find the CLAUDE.md you own in the current repo:

- **In `~/clawd/`** — update your agent file (e.g. `yuki/CLAUDE.md`, `kestrel/CLAUDE.md`)
- **In any other repo** — update the project root `CLAUDE.md`

If unsure, ask which file to update.

## How to Update

1. **Review the conversation** — identify decisions, preferences, patterns, or conventions that were established
2. **Read the current CLAUDE.md** — understand what's already documented
3. **Add only what's new** — don't duplicate existing content
4. **Place it in the right section** — match the file's existing structure. Create a new section only if nothing fits.
5. **Be concise** — write it as an instruction, not a narrative. Future agents need to act on it, not read a story.

## What to Capture

- Workflow decisions ("always use X instead of Y")
- Tool preferences ("use ttal pr, not gh")
- Conventions discovered ("tests go in __tests__/ not test/")
- Architecture decisions ("we chose X because Y")
- Gotchas and warnings ("don't run Z, it breaks W")

## What NOT to Capture

- Session-specific details (current task, in-progress work)
- Anything already in CLAUDE.md (check first!)
- Speculative conclusions from one conversation
- Temporary workarounds (unless they'll persist)

## After Updating

- Show the user what you added/changed
- Commit if asked (follow the repo's commit conventions)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.