agentleFS
Sign inStart free
danoszzdanoszz/penthera/AGENTS.md

penthera

AGENTS.md79 starsChanged 4 months agoMITView on GitHub

Project page. For coding agents asking "how do I work on Penthera, or use it, without breaking its rules?" Penthera is an offline, rule-based security agent (MIT, Node.js 22 or later). Deterministic rules decide every finding. A model may only propose. Tests must stay offline and parallel-safe: temp dirs, port 0, no real HOME. The details are in CONTRIBUTING.md. 1. Add a changelog fragment in changelog.d/ with a section: line (format). Do not edit CHANGELOG.md. If you touch the CLI,…

We split this AGENTS.md into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No version bumps or tags: Only the maintainer starts a release.

    line 51
  2. Distinctive:

    Rules decide: Do not add a code path where a model's output becomes a finding, a fix or an exit code without a deterministic check.

    line 52
  3. Very specific:

    Docs follow the [voice guide](docs/voice.md): Every command you document must exist in --help.

    line 58

≈ “Common” is an estimate: at least 1 in 5 of the 36,368 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md

Project page. For coding agents asking "how do I work on Penthera, or use it, without breaking its rules?"

Penthera is an offline, rule-based security agent (MIT, Node.js 22 or later). Deterministic rules decide every finding. A model may only propose.

Commands

CommandWhat it does
npm installInstall dependencies
npm testThe full suite (vitest), including docs-as-tests and the changelog check
npm run lintESLint: unused code, undefined names, unreachable code
npm run audit:internalThe release gate, written to audit/REPORT.md; includes the docs link check
npm run changelogChecks every fragment in changelog.d/

Tests must stay offline and parallel-safe: temp dirs, port 0, no real HOME. The details are in CONTRIBUTING.md.

Layout

bin/          CLI entry points
src/          CLI and orchestration: scan-url, scan-repo, reporter, ui, mcp, agent loop
lib/          the engine: blackbox, whitebox, deps, agents, install, hook, fix,
              advisories, rules, ai, memory, compliance
skills/       the Agent Skill (skills/penthera/SKILL.md)
tests/        vitest suites and fixtures
bench/        the benchmark and the E10 known-vulnerable-apps runner
docs/         user docs (start at docs/index.md); docs/research/ holds research notes
changelog.d/  one changelog fragment per pull request
site/         penthera.org

Rules

  1. Add a changelog fragment in changelog.d/ with a section: line (format). Do not edit CHANGELOG.md. If you touch the CLI, MCP tools, rules or the Action, update the doc that owns it, or write docs: none, <reason> in the fragment. CI checks it (scripts/ci/docs-gate.mjs).

  2. Add the AI-Assisted trailer to every commit you write. Do not add a Co-Authored-By: line. CI checks it (scripts/ci/provenance.mjs):

    AI-Assisted: <tool> with <model and version>; <what it did, in one line>
  3. No version bumps or tags. Only the maintainer starts a release.
  4. Rules decide. Do not add a code path where a model's output becomes a finding, a fix or an exit code without a deterministic check.
  5. No run-time downloads of rules, templates or signatures. Rules ship in releases.
  6. Only authorized targets. Scan your own code, loopback, or hosts the maintainer named. Never point a scan or a test at a third-party system.
  7. Docs follow the voice guide. Every command you document must exist in --help; npm test checks it.

Using Penthera from an agent

Pin the version. latest on npm is 2.1.2, with the MCP server, AI mode, library mode and the agent brief.

RouteCommandRead
CLInpx penthera@2.1.2 --repo .scan
Agent Skillnpx skills add danoszz/pentheraskill
MCP serverclaude mcp add penthera -- npx -y penthera@2.1.2 mcpmcp

The MCP tools are read-only, read only inside your project, never apply fixes and run no shell. scan_url is passive and reaches only loopback or hosts the user listed with --allow-host. A redirect may also move within the target's registrable domain.

Download now (ZIP)

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 28,693 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository2 other files it gives its agents.

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