# Generate Claude Md

> Generate a lean, high-signal `CLAUDE.md` for a repository by reusing the source workflow under `claude/generate-claude-md`. Use when the user wants a CLAUDE.md created or refreshed in Codex.

- **Type:** Skill
- **Install:** `agentstack add skill-smykla-skalski-sai-generate-claude-md`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [smykla-skalski](https://agentstack.voostack.com/s/smykla-skalski)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [smykla-skalski](https://github.com/smykla-skalski)
- **Source:** https://github.com/smykla-skalski/sai/tree/main/codex/generate-claude-md

## Install

```sh
agentstack add skill-smykla-skalski-sai-generate-claude-md
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Generate CLAUDE.md

Use this skill when the user wants a `CLAUDE.md` file created, scaffolded, or
refreshed for a repository from codebase analysis.

This is a thin Codex wrapper around
`claude/generate-claude-md/skills/generate-claude-md/SKILL.md`. Reuse the source
workflow, references, and validator from the Claude skill directory instead of
maintaining duplicated Codex copies.

## Use This Skill

- Use this skill when the user wants a `CLAUDE.md` generated or refreshed for a project.

## Do Not Use This Skill

- Do not use this skill to audit an existing CLAUDE.md against a checklist — that is `review-claude-md`.
- Do not use this skill for generic markdown authoring unrelated to CLAUDE.md.

## Source Material

- Source skill: `claude/generate-claude-md/skills/generate-claude-md/SKILL.md`
- Source directory: `claude/generate-claude-md/skills/generate-claude-md/`
- Codex metadata: `agents/openai.yaml`

Load only the source files needed for the current task. Do not recreate or copy the
Claude-side bundled resources into this Codex skill.

## Workflow

1. Treat `claude/generate-claude-md/skills/generate-claude-md/SKILL.md` as the source workflow.
2. Read the needed source references from `claude/generate-claude-md/skills/generate-claude-md/references/` (principles, section guide, scan spec, modularization, example).
3. Run the source validator `claude/generate-claude-md/skills/generate-claude-md/scripts/validate-claude-md.py` instead of duplicating it.
4. Apply the deletion test to every line; produce a short, project-specific file.

## Codex Notes

- Ignore Claude-only frontmatter and runtime wiring such as `allowed-tools`, `context: fork`, `$ARGUMENTS`, and `CLAUDE_SKILL_DIR`.
- Infer the repo root and flags from the user request and local context before asking follow-up questions.
- Write safety: never overwrite an existing `CLAUDE.md` unless the user was explicit; otherwise write `CLAUDE.generated.md` and report.
- If a source script fails because of sandbox restrictions, rerun it with escalation and a short justification.

## Verification

After generating, run the source validator on the written file and report the verdict.

1. Run `validate-claude-md.py  --human` and read the result.
2. Fix any `[FAIL]` check and re-run, up to a few rounds.
3. Report what was verified, the final line count, and any residual risk.

## Examples

User: "Use Generate CLAUDE.md for this repo."
Assistant: Opens `claude/generate-claude-md/skills/generate-claude-md/SKILL.md`, scans the codebase, synthesizes a lean CLAUDE.md, validates it, and reports.

User: "Make a CLAUDE.md for this project."
Assistant: Reuses the source material from `claude/generate-claude-md/skills/generate-claude-md/`, writes a short project-specific file, and runs the validator before reporting.

## Source & license

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

- **Author:** [smykla-skalski](https://github.com/smykla-skalski)
- **Source:** [smykla-skalski/sai](https://github.com/smykla-skalski/sai)
- **License:** MIT

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

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-smykla-skalski-sai-generate-claude-md
- Seller: https://agentstack.voostack.com/s/smykla-skalski
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.