Bootstrap project

Skill ada-ggf25/AI-Tools/global/codex/skills/bootstrap-project

A cross-device synced catalog for Claude and Codex skills, agents, and workflows.

Install

npx -y skills add ada-ggf25/AI-Tools --skill bootstrap-project

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

5 stars5 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

One-command Codex project setup orchestrator. Detects the stack, then runs root AGENTS.md setup, scoped AGENTS.md placement, Codex permission/config review, automation proposals, CI setup, and optional Obsidian Vault creation with approval gates. Global and project-agnostic. Trigger when the user says "bootstrap project", "bootstrap-project", "set up this repo for Codex", "onboard this project", or "run the full project setup".

SKILL.md

3.3 KB, as published. Nobody here has run it

Bootstrap A Project For Codex

Run the Codex-oriented new-repo setup workflow in one guided pass. This is a thin orchestrator: it owns sequencing and handoffs, and delegates detailed work to the focused skills.

Sequence:

AGENTS.md setup -> where-agents-md -> setup-permissions -> propose-automation -> setup-ci optional -> obsidian-vault optional -> durable memory suggestions.

Operating Principles

Orchestrate, do not reimplement delegated skills.
Pause after each step and continue only after user approval.
Be idempotent: detect existing AGENTS.md, .codex/, .agents/, and repo config.
Prefer shared repo guidance only when it is useful to teammates; keep personal Codex config personal.

Procedure

0. Orient

Confirm the working directory is the intended repo root.
Detect existing setup: root/nested AGENTS.md, AGENTS.override.md, .codex/, .agents/, CLAUDE.md, README, CI, and project manifests.
For a large repo, delegate broad exploration to a read-only subagent when available.
Show the planned sequence and let the user drop steps.

1. Root Codex Instructions

If no root AGENTS.md exists, propose creating one from observed repo facts.
If one exists, offer to review/update it instead of regenerating.
If CLAUDE.md exists, use it as source context but do not copy Claude-only commands, hook syntax, or permission syntax blindly.

2. Scoped Instruction Placement

Invoke where-agents-md to identify high-value nested AGENTS.md candidates.
Present recommendations and create files only for directories the user approves.

3. Codex Permissions And Config

Invoke setup-permissions to review Codex sandbox, trusted project, and command approval expectations.
Do not edit global ~/.codex/config.toml unless the user explicitly asks. Prefer project-scoped guidance or a patch proposal.

4. Automation Proposal

Invoke propose-automation to suggest project Codex skills, custom agents, hooks, or plugin packaging.
Scaffold only approved items.

5. CI Pipeline

Optionally invoke setup-ci for GitHub Actions. Skip if the repo is not on GitHub or if the user declines.

6. Obsidian Vault

Optionally invoke obsidian-vault if the user wants a repo knowledge base.

7. Durable Memories

Offer only genuinely durable, non-secret facts as project memories. Do not record information already captured in AGENTS.md or README.

8. Wrap Up

Summarize created/changed files.
Remind the user that new Codex skills and agents are picked up in a new Codex session.

Guardrails

Never run the full chain unattended.
Never clobber existing setup; merge or propose targeted edits.
Keep Claude and Codex artifacts separate unless the user explicitly asks for a shared compatibility file.
Never fabricate repo facts.