agentleFS
Sign inStart free
oomol-laboomol-lab/leina-agent/AGENTS.md

leina-agent

AGENTS.md4 starsChanged 31 days agoMITView on GitHub

This repository builds the standalone open-source Leina Agent Docker distribution from a pinned NousResearch Hermes Agent revision. These rules are mandatory. Read the document matching the work before editing: This is a Docker distribution project, not an independent Hermes fork. It may assemble OO CLI, curated Skills, provider Plugins, and minimal generic Hermes patches that cannot use an existing extension surface. Do not add OOMOL production bindings, internal operations tooling, messaging platform patches, or reimplementations of Hermes core behavior. Do…

  • Reads credentials
  • 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 31,542 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 repository3 other files it gives its agents.

The file

Show its rules and text

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

19 rules. About 2 are common, 8 are distinctive, and 9 are very specific to this project.

  • ≈2

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Read the document matching the work before editing:

    line 9
  2. Very specific:

    Do not add OOMOL production bindings, internal operations tooling, messaging platform patches, or reimplementations of Hermes core behavior.

    line 24
  3. Distinctive:

    Prefer the distribution layer: Docker, configuration, scripts, Skills, Plugins, tests, and public documentation.

    line 30
  4. Distinctive:

    Add an upstream patch only when no extension surface can implement the behavior.

    line 32
  5. Very specific:

    Treat upstream.lock.json as the canonical Hermes and OO CLI version record.

    line 35
  6. Very specific:

    Follow docs/upstream-maintenance.md for upstream changes and keep each upstream update in its own commit.

    line 38
  7. Distinctive:

    Keep Hermes's 60-character Skill-description default.

    line 47
  8. Distinctive:

    Verify artifacts created by document-producing Skills by reading or rendering them back.

    line 49
  9. Common:

    Never log prompts, uploaded contents, credentials, signed URLs, or sensitive raw responses.

    line 54
  10. Very specific:

    Keep non-secret configuration in config.yaml.

    line 56
  11. Very specific:

    Keep mutable state under /data.

    line 64
  12. Very specific:

    Run the final process as the hermes user and keep the entrypoint small and transparent.

    line 67
  13. Common:

    Do not add private registries, internal endpoints, or company credentials.

    line 69
  14. Very specific:

    Run repository tooling through Docker.

    line 73
  15. Very specific:

    Follow docs/development.md.

    line 75
  16. Distinctive:

    Run git diff --check before committing.

    line 77
  17. Distinctive:

    Keep unrelated changes in separate Conventional Commits.

    line 78
  18. Very specific:

    Never commit .env, credentials, user data, generated artifacts, runtime state, or downloaded Hermes source.

    line 83
  19. Distinctive:

    Use placeholders in public examples and document current public behavior, not internal deployment history.

    line 85

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 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

Leina Agent Development Guide

This repository builds the standalone open-source Leina Agent Docker distribution from a pinned NousResearch Hermes Agent revision. These rules are mandatory.

Required Reading

Read the document matching the work before editing:

WorkRequired document
Project boundaries and runtime layoutdocs/architecture.md
Local setup, tests, or Docker buildsdocs/development.md
Hermes version or patch changesdocs/upstream-maintenance.md
Current migration status and next workdocs/handoff.md

Project Boundaries

This is a Docker distribution project, not an independent Hermes fork. It may assemble OO CLI, curated Skills, provider Plugins, and minimal generic Hermes patches that cannot use an existing extension surface.

Do not add OOMOL production bindings, internal operations tooling, messaging platform patches, or reimplementations of Hermes core behavior. Do not copy internal operational code from oomol/hermes-agent merely because it exists.

Change Policy

  • Prefer the distribution layer: Docker, configuration, scripts, Skills, Plugins, tests, and public documentation.
  • Add an upstream patch only when no extension surface can implement the behavior. Keep patches small, reviewable, and fail closed with git apply --check; never replace a complete upstream source file.
  • Treat upstream.lock.json as the canonical Hermes and OO CLI version record. Pin Hermes by full commit SHA and OO CLI artifacts by version and per-arch SHA-256.
  • Follow docs/upstream-maintenance.md for upstream changes and keep each upstream update in its own commit.

Skills And Providers

  • Install only Skills listed in config/curated-skills.txt or config/hermes-skills.txt. Reject collisions, traversal, and symlinks.
  • Install runtime dependencies at image build time. Keep image-owned Skills immutable and user Skills under $HERMES_HOME.
  • Keep Hermes's 60-character Skill-description default. Any larger limit must be an explicit per-Skill exception.
  • Verify artifacts created by document-producing Skills by reading or rendering them back.
  • Register providers through Hermes provider ABCs; do not add model tools to core.
  • Invoke OO CLI with argument arrays, bound timeouts, and sanitized errors. Never log prompts, uploaded contents, credentials, signed URLs, or sensitive raw responses.
  • Keep non-secret configuration in config.yaml. OO credentials come only from the required runtime OO_API_KEY, never image layers or persisted files.
  • Provider failures must not prevent Hermes from starting.

Runtime And Security

  • Startup requires OO_API_KEY, OO_LLM_BASE_URL, OO_LLM_MODEL, and OO_LLM_API_MODE, but it must not validate them with a network request.
  • Keep mutable state under /data; $HERMES_HOME and OO_CONFIG_DIR must live below it.
  • Seed config.yaml only when missing and never overwrite user changes.
  • Run the final process as the hermes user and keep the entrypoint small and transparent.
  • Do not add private registries, internal endpoints, or company credentials.

Validation And Commits

  • Run repository tooling through Docker. uv may be used inside build containers but must not be a host prerequisite.
  • Follow docs/development.md; release and upstream changes require the full Docker integration test.
  • Run git diff --check before committing.
  • Keep unrelated changes in separate Conventional Commits.

Open-Source Hygiene

  • Preserve the OOMOL MIT license and required upstream or third-party notices.
  • Never commit .env, credentials, user data, generated artifacts, runtime state, or downloaded Hermes source.
  • Use placeholders in public examples and document current public behavior, not internal deployment history.

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