Borrowing it
Nothing to install: this file belongs to matt-w-horn/skills. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/matt-w-horn/skills/main/CLAUDE.mdgit clone --depth 1 https://github.com/matt-w-horn/skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/matt-w-horn/skills/claude-md)<a href="https://agentmods.dev/instructions/matt-w-horn/skills/claude-md"><img src="https://agentmods.dev/badge/instructions/matt-w-horn/skills/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/instructions/matt-w-horn/skills/claude-md"><img src="https://agentmods.dev/badge/instructions/matt-w-horn/skills/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.01128 | $0.01128 |
| Opus 5 | $0.00564 | $0.00564 |
| Sonnet 5 | $0.00226 | $0.00226 |
| Haiku 4.5 | $0.00113 | $0.00113 |
Grade A, and why
skills CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Guidance for Claude Code working in this repository.
What this is
Personal Claude skills, one per directory under skills/. The repo is also a
Claude Code plugin marketplace (.claude-plugin/marketplace.json +
plugin.json), so others can /plugin marketplace add matt-w-horn/skills; I
load them myself by symlinking each into ~/.claude/skills/. A skill is a
directory with a SKILL.md (frontmatter: name matching the directory,
description) plus optional references/, scripts/, and tests/. The
simulation scripts (skills/life-paths/scripts/fi_model.py,
skills/financial-planning/scripts/simcore.py) are pure standard library;
keep them that way, since CI installs nothing.
Commands
python3 tools/validate_skills.py # structure + path-reference validation
tools/run_tests.sh # every tests/ suite (skills and tools)
sh tools/install-hooks.sh # git pre-commit hook: both checks + gitleaks
Both checks run on commit (hook) and on push/PR (CI). Keep them green.
Evals (see tools/eval/README.md; only the first is gated by CI, via
tools/eval/tests/):
python3 tools/eval/lint_evals.py # corpus quality gate, stdlib only
python3 tools/eval/grade.py --calibrate # judge must separate good.md from bad.md
python3 tools/eval/run_trigger.py --split graded # ~144 `claude -p` runs
python3 tools/eval/run_exec.py --runs 2 # 32 long runs; hours
Conventions & gotchas
- Adding a skill: create
skills/<skill>/with aSKILL.md, then activate it from the repo root withln -s "$PWD/skills/<skill>" ~/.claude/skills/<skill>. Nothing else to wire up: the marketplace auto-discovers everything underskills/(no manifest edit), and the validator finds skills at any depth by theirSKILL.md. - Path references are validated. Any
dir/filetoken in a skill's Markdown whose first segment is a real subdirectory of that skill must resolve, fenced commands included; runtime artifacts (dirs that don't exist in the repo, likefinances/config.json) are ignored. Files underreferences/orscripts/that no Markdown mentions draw a warning. - Scripts must ship with tests (
tests/test_*.py,unittest, run from the skill directory). Degenerate inputs should raiseValueErrorwith a clear message, never crash deep in the math — see the guards insimcore.pyandfi_model.pyfor the pattern. - Eval corpora live in
skills/<skill>/evals/and are authored blind: a trigger query or rubric assertion must not paraphrase the SKILL.md body, andlint_evals.pyrule 5 fails the corpus if one shares a five-word phrase with the body that is absent from the description. When editing evals, work from the description and domain knowledge, not the skill's own instructions. Trigger queries must also be self-contained: runs happen in an empty directory, so a query naming a file misfires for a fixture reason that reads like a description failure. - Never put a
SKILL.mdunderevals/.validate_skills.pytreats anySKILL.mdat any depth as a skill and would try to validate the fixture. Same trap forscripts/-,references/-,tests/- orevals/-prefixed path tokens in eval prose: the validator resolves them against the real directories. Both are linted. - Never set
versioninplugin.jsonor a marketplace plugin entry. Claude Code resolves a plugin's version fromplugin.json, then the marketplace entry, then the source commit SHA, and skips the update when the resolved version matches what a user already has. A literal version nobody remembers to bump freezes every existing install however many commits land, and it fails silently — nothing local can tell. With the field omitted, each commit is its own version.tools/tests/test_plugin_manifests.pyenforces this. - This repo is public, so it may not depend on anything outside itself. A
skill here must stand alone: no references to system-wide skills, to
~/.claude, or to a project that isn't public. Skills needing any of those live as real directories under~/.claude/skills/instead, which is whereapps-script-deploywent on 2026-07-26. The~/.claude/skills/paths below are install instructions for a reader, not a dependency. - No secrets belong here, ever; gitleaks runs in the hook and CI as a backstop.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 80 lines · 1,128 tokens per session scan A 2225256d797a
skills CLAUDE.md is an instructions file published in the GitHub repository matt-w-horn/skills (6 stars, last pushed 1mo ago), licensed MIT. It adds 1,128 tokens to every session, about $0.0056 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.