agentleFS
Sign inStart free
poizoncodedpoizoncoded/skills/AGENTS.md

skills

AGENTS.md2 starsChanged 4 months agoMITView on GitHub

This repo is the reusable PoizonCoded skill-set for future projects. Keep changes portable, private-data-free, and grounded in the actual .codex/ and .claude/ assets in this repository. - .codex/config.toml: Codex MCP and multi-agent settings. - .codex/skills/: Codex-discoverable skills and OpenAI UI metadata. - .claude/settings.json: Claude Code permissions and enabled MCP servers. - .claude/settings.local.json: local Claude Code settings. - .claude/skills/: Claude Code skills shared with the Codex skill root. - .claude/commands/, .claude/agents/, .claude/helpers/: Claude Code commands, agents, hooks, statusline, and helper scripts.…

The file

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

10 rules. About 1 is common, 4 are distinctive, and 5 are very specific to this project.

  • ≈1

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When working on multi-file tasks or complex features, use ToolSearch to look for Ruflo MCP tools first.

    line 34
  2. Distinctive:

    If Ruflo tools are not exposed in the session, continue with local repository tools and keep the Ruflo configuration intact.

    line 37
  3. Distinctive:

    Every skill directory under either root must have a matching directory under the other root.

    line 45
  4. Very specific:

    Use rg / rg --files for search.

    line 116
  5. Very specific:

    Use apply_patch for manual file edits.

    line 117
  6. Common:

    Keep the bundle reusable; avoid project-specific app names, ports, private paths, credentials, resumes, customer data, or generated …

    line 118
  7. Distinctive:

    Prefer small, explicit skill descriptions that start with "Use when..." and include the real trigger terms future agents should search for.

    line 120
  8. Very specific:

    Keep agents/openai.yaml values short and consistent with the corresponding SKILL.md.

    line 122
  9. Very specific:

    When adding a shared skill, create or update both roots and update this guide plus .claude/settings.json permissions if explicit skill …

    line 124
  10. Distinctive:

    Prefer the narrowest relevant check first, then broaden before claiming a fix is complete:

    line 130

≈ “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

poizoncoded - Agent Guide

This repo is the reusable PoizonCoded skill-set for future projects. Keep changes portable, private-data-free, and grounded in the actual .codex/ and .claude/ assets in this repository.

Project Map

  • .codex/config.toml: Codex MCP and multi-agent settings.
  • .codex/skills/: Codex-discoverable skills and OpenAI UI metadata.
  • .claude/settings.json: Claude Code permissions and enabled MCP servers.
  • .claude/settings.local.json: local Claude Code settings.
  • .claude/skills/: Claude Code skills shared with the Codex skill root.
  • .claude/commands/, .claude/agents/, .claude/helpers/: Claude Code commands, agents, hooks, statusline, and helper scripts.
  • .agents/, .swarm/, .superpowers/: runtime placeholders for agent workflows.
  • scripts/init-workspace.mjs: npx/bootstrap installer for new repositories, including --ai codex|claude|both and --profile fe|be|bot|all selectors.
  • package.json: skill-set identity, npx bin, and Ruflo integration metadata.

Main Harness

ruvnet/ruflo is the primary agent harness for this skill-set. Prefer Ruflo-backed MCP tools, memory, hooks, and swarm orchestration when they are available.

Codex starts the Ruflo MCP server through .codex/config.toml:

npx -y ruflo@latest mcp start

When working on multi-file tasks or complex features, use ToolSearch to look for Ruflo MCP tools first. Start with hooks_route when available; use swarm_init / agent_spawn only when independent workers are genuinely useful. If Ruflo tools are not exposed in the session, continue with local repository tools and keep the Ruflo configuration intact. The bootstrap CLI should not install Ruflo just to copy assets; target repositories can opt into a local published ruflo dependency with --with-ruflo-dep.

Skills

Project skills are shared under both .codex/skills/ and .claude/skills/. Every skill directory under either root must have a matching directory under the other root. When changing a skill, update both roots together and keep content in sync, except for Codex-only agents/openai.yaml adapters and upstream package metadata such as README.md or metadata.json.

  • .codex/skills/poizoncoded-conventions and .claude/skills/poizoncoded-conventions: use before changing this skill-set repo or adapting it for a new workspace.
  • .codex/skills/feature-sliced-design and .claude/skills/feature-sliced-design: use for UI file placement and import boundaries.
  • .codex/skills/react-best-practices and .claude/skills/react-best-practices: use for React component, hook, data fetching, and rendering/performance work.
  • .codex/skills/verify-md and .claude/skills/verify-md: use when verifying or rewriting markdown docs in this repo or a project derived from it.
  • .codex/skills/dual-tool-skills and .claude/skills/dual-tool-skills: use when adding/changing project skills or doc references to skills.
  • .codex/skills/golang-* and .claude/skills/golang-*: use for Go project layout, style, testing, concurrency, database, CLI, observability, performance, security, and common library work.
  • .codex/skills/qdrant-* and .claude/skills/qdrant-*: use for Qdrant client integration, deployment, monitoring, scaling, model migration, performance, search quality, and version upgrades.
  • .codex/skills/agentdb-* and .claude/skills/agentdb-*: use for AgentDB memory, vector search, learning, optimization, and advanced coordination work.
  • .codex/skills/github-* and .claude/skills/github-*: use for GitHub code review, multi-repo, project management, release, and workflow automation work.
  • .codex/skills/reasoningbank-* and .claude/skills/reasoningbank-*: use for ReasoningBank memory and adaptive intelligence workflows.
  • .codex/skills/swarm-* and .claude/skills/swarm-*: use for Ruflo or agentic-flow swarm orchestration and advanced swarm work.
  • .codex/skills/v3-* and .claude/skills/v3-*: use for claude-flow v3 implementation, integration, memory, MCP, security, performance, and swarm coordination work.
  • .codex/skills/nest and .claude/skills/nest: use for generic NestJS backend module, provider, controller, and bootstrap work.
  • .codex/skills/typeorm and .claude/skills/typeorm: use for generic TypeORM entities, repositories, query builders, relations, migrations, and database configuration.
  • .codex/skills/browser and .claude/skills/browser: use for browser automation guidance.
  • .codex/skills/hooks-automation and .claude/skills/hooks-automation: use for Claude/Ruflo hook automation workflows.
  • .codex/skills/pair-programming and .claude/skills/pair-programming: use for structured pair-programming workflows.
  • .codex/skills/skill-builder and .claude/skills/skill-builder: use for building new skills.
  • .codex/skills/sparc-methodology and .claude/skills/sparc-methodology: use for SPARC workflow guidance.
  • .codex/skills/stream-chain and .claude/skills/stream-chain: use for stream-chain workflow guidance.
  • .codex/skills/verification-quality and .claude/skills/verification-quality: use for verification and quality assurance workflows.
  • The repo vendors mcollina/skills project skills under both roots at upstream commit 879c9deb0d3a3551086e7c737577dc6cd76fd0c7 (2026-07-16): documentation, fastify, init, linting-neostandard-eslint9, node, nodejs-core, oauth, octocat, skill-optimizer, snipgrapher, and typescript-magician.

Claude-focused or Ruflo-focused skills may still mention Claude Code, claude-flow, or MCP-specific commands, but the skill folders themselves stay shared so both agents see the same guidance.

Editing Rules

  • Preserve user changes. Never revert unrelated dirty work.
  • Use rg / rg --files for search.
  • Use apply_patch for manual file edits.
  • Keep the bundle reusable; avoid project-specific app names, ports, private paths, credentials, resumes, customer data, or generated personal content.
  • Prefer small, explicit skill descriptions that start with "Use when..." and include the real trigger terms future agents should search for.
  • Keep agents/openai.yaml values short and consistent with the corresponding SKILL.md.
  • When adding a shared skill, create or update both roots and update this guide plus .claude/settings.json permissions if explicit skill invocation should be allowed there.

Verification

Prefer the narrowest relevant check first, then broaden before claiming a fix is complete:

npm run check:json
npm run check:skills
git diff --check

For reference hygiene, scan active docs/config for stale project language:

npm run check:refs

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