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/backendgamesetmatch/sourcebook/claude-mdgit clone --depth 1 https://github.com/BackendGameSetMatch/sourcebookWhat 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 | $0.00471 | $0.00471 |
| Opus 5 | $0.00235 | $0.00235 |
| Sonnet 5 | $0.00094 | $0.00094 |
| Haiku 4.5 | $0.00047 | $0.00047 |
Grade A, and why
sourcebook 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 — 35 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI Documentation
This project maintains structured internal documentation in /docs/ai/.
When to consult docs
- Before modifying core service or library code - search for the system or module
- Before working on this tool's own design - load the
designbundle (the only bundle this repo defines) - Before touching release or deployment workflows - read the relevant process doc
- Before proposing any architectural approach - check
decisions/for existing ADRs - When encountering unfamiliar internal terms - check the glossary
- When unsure about constraints - read
conventions.md(these are hard rules, not suggestions)
How to access
Use the sourcebook-lookup skill: /sourcebook-lookup search <keywords>
Or load a domain bundle: /sourcebook-bundle <scope>. This repo defines one bundle, design;
passing an unknown scope prints the list, so check there rather than guessing a name.
Keep docs alive
- When you read a doc and notice it's wrong or incomplete, fix it with
update_document - When work you just did confirms a doc still matches reality, call
verify_document— one call, bumpslast_verifiedwith your attribution. Stale docs served byread_documentcarry a notice asking for exactly this. - When you learn something new about a system while helping a developer, update the relevant doc
- When a developer mentions a term not in the glossary, add it with
add_glossary_term - When you answer a question by synthesizing info from multiple docs, consider whether a new doc or bundle would help next time
- Run
lint_docsperiodically to find and fix structural issues - Always pass
authorwhen callingupdate_document,add_glossary_term, oringest_source. Get the value from the current git user: rungit config user.nameandgit config user.email, then format as"Name <email>". This gets added as Co-Authored-By in the git commit so we can trace who triggered each change.
Hard rules from conventions
Never violate constraints listed in /docs/ai/conventions.md.
Check decisions before proposing architectural changes.
Flag stale docs (check last_verified dates) rather than trusting outdated information.
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 · 35 lines · 471 tokens per session scan A 75f25061a58e
sourcebook CLAUDE.md is an instructions file published in the GitHub repository BackendGameSetMatch/sourcebook (3 stars, last pushed 21d ago), licensed MIT. It adds 471 tokens to every session, about $0.0024 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
neo AGENTS.md
AGENTS.md instructions for neomjs/neo, covering ai agent per-turn operational mandates, §corevalues, §identitypromptfirewall, §criticalgates and §precommitgates.
agentic-context-engine AGENTS.md
Instructions for kayba-ai/agentic-context-engine, covering agents.md, repository guidelines, pipeline-first development (mandatory), core code protection and documentation maintenance.
activity-frames AGENTS.md
Instructions for nossa-y/activity-frames, covering can i install it without asking?, the mental model (10 seconds), fastest integration: mcp, from python and things you should not do.
keep-the-why AGENTS.md
Instructions for oliver-zehentleitner/keep-the-why, covering agents.md and working conventions (no why needed, just follow these).
awesome-harness-engineering AGENTS.md
AGENTS.md instructions for ai-boost/awesome-harness-engineering, covering agents.md, what this repo is, conventions, what belongs and what doesn't belong.
Meterless AGENTS.md
Instructions for Meterless/Meterless, covering agent instructions for meterless, routing table, isolation rule, repo-wide rules and non-coding agents.