Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/cloudroad-io/skillcraft/claude-mdgit clone --depth 1 https://github.com/cloudroad-io/skillcraftWhat 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 | $0.00919 | $0.00919 |
| Opus 5 | $0.00460 | $0.00460 |
| Sonnet 5 | $0.00184 | $0.00184 |
| Haiku 4.5 | $0.00092 | $0.00092 |
Grade A, and why
skillcraft 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 2d 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 — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
skillcraft
ESLint + Jest for agent-config files. skillcraft lints, syncs and scaffolds the fragmented ecosystem of SKILL.md, CLAUDE.md, AGENTS.md, .cursor/rules, .claude/rules and copilot-instructions. One canonical source, many managed targets, drift detection in CI.
Install
uv tool install skillcraft
# or: pip install skillcraft
Commands
| Command | Purpose |
|---|---|
skillcraft lint [--check] [-f plain|json|github|sarif] |
Run the rule set over discovered config files; exit 1 on any ERROR. |
skillcraft sync [--check] [--diff] [--adopt <file>] |
Regenerate managed targets from AGENTS.md; detect drift. |
skillcraft init [--name <name>] |
Scaffold a minimal AGENTS.md + .skillcraft.toml. |
skillcraft version |
Print the version. |
How it works
- Canonical source =
AGENTS.md(vendor-neutral, schema-less). Richer metadata (name, description, scope, license, …) rides in invisible<!-- skillcraft:meta <json> -->comments — valid markdown to every consumer, machine-readable toskillcraft. - IR:
ConfigDoc. Every parser emits it, every renderer consumes it. Same-format parse→render is lossless;extra_frontmatterescape hatch guarantees no field is silently dropped. - Sync.
skillcraft syncrenders each target from the canonical doc and writes it with a<!-- skillcraft:managed-source path=AGENTS.md -->marker.skillcraft sync --checkexits 1 if any managed target drifted (CI). Unmanaged files are never overwritten; opt in with--adopt. - Plugins. Subclass
RuleorConverter, decorate with@register_rule/@register_converter, and (for external packages) declare an entry-point inskillcraft.rules/skillcraft.converters. SeeCONTRIBUTING.md.
Rules
| ID | Scope | Rule |
|---|---|---|
| SC101 | SKILL | name is kebab-case, ≤64 chars |
| SC102 | SKILL | in a skills/<name>/ folder, name matches the folder |
| SC103 | SKILL | description present, ≤1024 chars |
| SC104 | SKILL | body ≈ <5000 tokens (warn past 4000) |
| SC105 | SKILL | description ≥40 chars for triggerability (warn) |
| SC201 | CLAUDE | @path imports resolve, no cycles, ≤4 hops |
| SC202 | CLAUDE | line count <200 (warn), <500 (error) |
| SC203 | CLAUDE | @imports resolve inside the repo root (error) |
| SC204 | ALL | no skipped heading levels (warn) |
| SC301 | ALL | required frontmatter present iff the format requires it |
| SC302 | ALL | no merge-conflict markers in the body |
| SC304 | ALL | body ends with a trailing newline (warn) |
| SC401 | CURSOR | globs well-formed and the rule is reachable (error/warn) |
| SC402 | CURSOR | not both alwaysApply: true and globs (warn) |
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.
- 2d ago First seen · 71 lines · 919 tokens per session scan A 367f0c3730aa
skillcraft CLAUDE.md is an instructions file published in the GitHub repository cloudroad-io/skillcraft (2 stars, last pushed 1mo ago), licensed MIT. It adds 919 tokens to every session, about $0.0046 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
ai-agent-instruction-modules AGENTS.md
Instructions for tablackburn/ai-agent-instruction-modules, covering ai agent instructions, instructions for ai agents, instruction applicability matrix, available instruction files and quick reference.
tokrepo-search-skill AGENTS.md
Instructions for henu-wang/tokrepo-search-skill, covering tokrepo — ai asset registry, search, install, fetch raw content and mcp server.
ai-agent-instruction-modules CLAUDE.md
Instructions for tablackburn/ai-agent-instruction-modules, a project described as: AIM: Modular, opt-in AI agent instructions for any codebase.
memorix GEMINI.md
Instructions for AVIDS2/memorix, covering memorix - cross-agent memory rules, session start - bind project, then load context, during session - capture important context, architecture & decisions and bug fixes & problem solving.
Ctxo CLAUDE.md
Instructions for alperhankendi/Ctxo, covering claude.md — ctxo, project overview, quick reference, dev (pnpm workspace) and usage (consumer).
awesome-reviewers CLAUDE.md
Instructions for baz-scm/awesome-reviewers, covering awesome reviewers — repository guidelines, what this project is, source of truth, site layer and machine interface.