claude_skills / rules
AGENTS.md67 starsChanged 50 days agoMITView on GitHub
rules/ is the authoritative, plain-content rule source for tools without a native path-glob mechanism (Codex, Hermes). manifest.json is the authoritative source for which glob patterns load which file — read it directly rather than trusting a restated pattern here. context-loader.mjs is designed as the shared matcher/loader for every harness's own hook wrapper to call; as of this writing only Claude Code's wrapper (.claude/hooks/context-rules.mjs) exists and is wired (.claude/settings.json's PostToolUse/SessionStart hooks) — a Codex or Hermes wrapper has not been added…
In this repository41 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- research/CLAUDE.mdCLAUDE.md
- .cursor/rules/backlog-before-work.mdcCursor rule
- agent-browserSkill
- beadsSkill
- cccSkill
- copy-editingSkill
- daily-releasesSkill
- graph-engineeringSkill
- rebaseSkill
- receiving-pr-reviewsSkill
- topic-specialistSkill
- workshop-question-framingSkill
- agent-browserSkill
- boilSkill
- commit-stagedSkill
- complete-milestoneSkill
- cove-prompt-designSkill
- create-merge-request-changelogSkill
- create-milestoneSkill
- design-anti-patternsSkill
- evaluate-sdlc-layersSkill
- example-argument-substitutionSkill
- external-pattern-integratorSkill
- fact-checkSkill
- ghSkill
- git-history-reconSkill
- group-items-to-milestoneSkill
- knowledge-explorerSkill
- linear-walkthroughSkill
- modern-gitSkill
- prepare-walkthrough-presentationSkill
- readme-badgerSkill
- refresh-researchSkill
- research-curatorSkill
- session-historianSkill
- seven-prompt-content-engineSkill
- skill-research-processSkill
- start-milestoneSkill
- universe-of-thoughtsSkill
The file
Files
rules/ — Cross-Tool Path-Scoped Rules
rules/ is the authoritative, plain-content rule source for tools without a native path-glob mechanism
(Codex, Hermes). manifest.json is the authoritative source for which glob patterns load which
file — read it directly rather than trusting a restated pattern here. context-loader.mjs is
designed as the shared matcher/loader for every harness's own hook wrapper to call; as of this
writing only Claude Code's wrapper (.claude/hooks/context-rules.mjs) exists and is wired
(.claude/settings.json's PostToolUse/SessionStart hooks) — a Codex or Hermes wrapper has not
been added yet.
| File | Purpose | Reaches an agent via |
adr-lifecycle.md | ARCHITECTURE.md defines desired state; backlog records gaps; an ADR records a dismissible deliberation that nothing links to | manifest.json |
agent-output-contracts.md | prohibited silent-output instructions and enforcement checklist | manifest.json |
astral-tool-overrides.md | this repo's uv/ty/ruff policy overrides | manifest.json |
ci-workflows.md | CI workflow modification protocol | manifest.json |
citation-requirements.md | factual claims need a cited source | manifest.json |
data-format-selection.md | pick the format closest to the consumer; never parse prose for your own data | manifest.json |
delegation-format.md | wrong delegation-instruction formats to avoid in prose | manifest.json |
exception-handling.md | narrow exception catches only, no broad except Exception | manifest.json |
frontmatter-requirements.md | required frontmatter fields | manifest.json |
language-conventions.md | language choice, .cjs/.mjs Node convention | manifest.json |
linting-exceptions.md | when a lint suppression is (rarely) acceptable | manifest.json |
markdown-file-references.md | code fence and markdown link conventions | manifest.json |
plugin-development.md | auto-discovery, versioning, local testing | manifest.json |
plugin-json.md | manifest schema requirements | manifest.json |
prose-file-classification.md | review-treatment decision tree for prose files | manifest.json |
python-development.md | PEP 723 scripts, no uv workspace, ty errors, repo overrides on the Python skills | manifest.json + routing line |
review-and-correction-discipline.md | structural vs content review gates, and what belongs in AGENTS.md | manifest.json |
runtime-vs-design-time.md | runtime vs. design-time audience, and a portable artifact's actual (installed) environment vs. its authoring repo | manifest.json |
script-invocation.md | shebang/execute-bit, run scripts directly | manifest.json |
silent-failure-prevention.md | write operations must report what changed | manifest.json |
skill-documentation-verification.md | skill docs are AI-facing, not user-facing | manifest.json |
shared-process-extraction.md | writing into an agent what another also needs, or finding the same process in two agents: one skill holds it, both load it | manifest.json |
skill-substitution.md | load-time string substitution gotcha | manifest.json |
uv-run-required.md | stop and tell the user when uv is unavailable — no pip/poetry/pipx/python fallback exists | manifest.json |
yaml-toml-libraries.md | ruamel.yaml/tomlkit only, never pyyaml | manifest.json |
commit-cadence-and-worktrees.md | small scoped commits, worktrees for concurrent writes | routing line |
delegation.md | substantive work is delegated; pointer to agent-orchestration:delegate | routing line + .claude/hooks/context-rules-agent.mjs |
fact-verification-first.md | WebSearch before planning around a named product/version | routing line + .claude/hooks/context-rules-prompt.mjs |
fix-delegation-discipline.md | reproduction-first cycle for bug-fix delegation | routing line + .claude/hooks/context-rules-agent.mjs |
interactive-terminal-workarounds.md | PTY providers when a tool needs a TTY | routing line |
large-file-write-strategy.md | skeleton+edit-fill above 25K chars | routing line |
model-selection.md | model/effort tier by cognitive requirement | routing line + .claude/hooks/context-rules-agent.mjs |
proactive-fix-gate.md | gate before acting on a self-discovered problem | routing line |
scratch-directory.md | .tmp/scratch/ fallback output convention | routing line |
A manifest.json entry means .claude/hooks/context-rules.mjs loads the file when a touched path matches its
glob — read the actual pattern in manifest.json itself, not restated here. "routing line" means
the rule carries no manifest.json entry; instead, a line under root AGENTS.md's "Situational
Rule Triggers" section names the file and the condition for reading it. A routing-line rule marked
with a hook filename also reaches an agent directly on that hook's event, independent of whether
the routing line gets read.
Writing a Rule File
Rules are the current requirement only — never provenance, citations, or narrative. Put those in
the commit message or PR description; put a durable architecture decision in docs/ instead.
Rules are read only when small. Tightening an existing rule means rewriting it from scratch as
flat directives, not Edit-trimming words from its existing structure.
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
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.