agentleFS
Sign inStart free
dvnc-labsdvnc-labs/agent-input-firewall/AGENTS.md

agent-input-firewall

AGENTS.md0 starsChanged 5 months agoMITView on GitHub

This repo is skill-starter, the omidsaffari Labs scaffold for shipping ONE small, forkable skill (or a single CLAUDE.md) as an authoritative OSS repo. Codex clones it with gh repo create omidsaffari/<slug> --template omidsaffari/skill-starter, then fills it in per this spec. Read this fully before editing. - Build the narrowest perimeter artifact that makes an existing agent (Claude Code / Cursor / ChatGPT) better. One job per skill. - Never frame as "AI-powered". Lead with the concrete problem it kills, or…

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 2

    1. dvnc-labs/type-led-launch-design★ 1100% same
    2. dvnc-labs/agent-config-audit★ 0100% same

    Edited · 0

    None found.

    Wrote their own version · 0

    None found.

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repositoryOne other file it gives its agents.

The file

Create a free account to mount
Files
AGENTS.md

AGENTS.md — build spec for a skill / CLAUDE.md artifact

This repo is skill-starter, the omidsaffari Labs scaffold for shipping ONE small, forkable skill (or a single CLAUDE.md) as an authoritative OSS repo. Codex clones it with gh repo create omidsaffari/<slug> --template omidsaffari/skill-starter, then fills it in per this spec. Read this fully before editing.

Doctrine (non-negotiable)

  • Build the narrowest perimeter artifact that makes an existing agent (Claude Code / Cursor / ChatGPT) better. One job per skill.
  • Never frame as "AI-powered". Lead with the concrete problem it kills, or "the open-source alternative to <incumbent>".
  • The metric is forks / awesome-list inclusion / citations, not stars. The README + the skill quality are the entire product.
  • Banned: general agent frameworks, chatbot wrappers, Lovable/v0/bolt clones.

What you edit vs leave

  • Edit: SKILL.md (the skill itself), README.md (the landing page), optional scripts/ references/ assets/, and the repo name/description.
  • Keep: the name + description frontmatter keys in SKILL.md (CI checks them), the .github/workflows/ci.yml gate, the MIT LICENSE (set the year + "Omid Saffari" — already done).
  • A CLAUDE.md artifact instead of a skill? Replace SKILL.md with a single top-level CLAUDE.md, drop the skill install section, and keep everything else.

Fill SKILL.md

  • name: lowercase-kebab, matches the repo slug.
  • description: front-load trigger words; say when it SHOULD and SHOULD NOT fire. Progressive disclosure shows only name+description until the skill loads, and long ones get shortened — so the first clause must carry the trigger.
  • Body: imperative steps, explicit inputs/outputs. Instructions over scripts unless deterministic behavior or external tooling is required.

Fill README.md (the conversion surface)

Above-the-fold order, in this order: title → one-line problem statement → demo GIF (assets/demo.gif, the single best 5–15s moment) → one badge row → install (Claude Code + Codex). Keep it skimmable; deep docs go in references/.

Ship checklist (do every release)

  1. First fill step: rm .skill-starter-template (this switches on the strict release CI gates and signals all placeholders must now be replaced).
  2. name/description valid; CI green.
  3. README.md filled; assets/demo.gif present; assets/og.png (1280×640, solid bg, <1MB) set as the repo's social-preview image.
  4. Repo is a template repo; MIT LICENSE; 10–20 exact-match Topics (e.g. skill, claude-code, codex, agent, + the niche).
  5. A GitHub Release + a one-line CHANGELOG entry.
  6. Distribution: submit to the relevant awesome-list(s); draft the build-log for the Labs catalog; one fresh Show HN (Sunday ~noon UTC, problem-framed title, repo in URL, founder comment in 5 min).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free