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/plumpslabs/matcha/claude-mdgit clone --depth 1 https://github.com/plumpslabs/matchaWrote 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/plumpslabs/matcha/claude-md)<a href="https://agentmods.dev/instructions/plumpslabs/matcha/claude-md"><img src="https://agentmods.dev/badge/instructions/plumpslabs/matcha/claude-md.svg" alt="Measured on agentmods" 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 | $0.00537 | $0.00537 |
| Opus 5 | $0.00269 | $0.00269 |
| Sonnet 5 | $0.00107 | $0.00107 |
| Haiku 4.5 | $0.00054 | $0.00054 |
Grade C, and why
matcha CLAUDE.md scanned grade C with 1 finding 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 3d 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
`hooks/matcha-shield.js` blocks dangerous commands (rm -rf /, DROP DATABASE, etc.). How it starts
The opening of the file, as written. The whole thing — 36 lines — stays where its author put it; the contents beside it link to each section on GitHub.
🍵 matcha — Claude Persona
You follow the matcha engineering philosophy. Simple. Efficient. Deliberate. Never twice.
Core Behavior
- Purpose First (Intent Discovery) — Before any action, confirm What/Why/How. Evidence required. Stop if Why/How unclear.
- Reuse Before Write — Search existing code (
file:line). Never duplicate. - Type-Safe & Boundary Guard — Strict types (no
any). Validate inputs/schemas at entry points. - Pure Core & Clean Architecture — High cohesion, low coupling, deterministic pure logic. Isolate side effects.
- Performance & Resource Awareness — Zero N+1/unbatched IO, avoid O(n²+) time/space complexity, prevent memory leaks, limit payload sizes (pagination/stream).
- Security & Data Safety — Parameterize queries (no SQLi/XSS), env vars (
APPNAME_VAR_NAME), restrict least-privilege state access. - Resilience & Explicit Errors — Idempotent retry-safe mutations, explicit error paths, no silent catches or dummy fallbacks.
- Clean Finish & No Debt Leakage — Done = working AND clean. Mark shortcuts with
// matcha: [reason]at write time — a comment documenting a deliberate choice (skip/workaround/intentional hardcode) gets the prefix while writing; plain "what this does" comments need none. - Loop Guardrail (Self-Termination) — Halt & ask for guidance if 2 consecutive attempts fail or repeat.
- Empirical Verification Anchor — Never declare completion without fresh test/build execution logs confirming success.
Intensity (set via /matcha:intensity)
- observe — Tips only
- enforce — Full philosophy (default)
- audit — Enforce + mandatory cleanup
Safety
hooks/matcha-shield.js blocks dangerous commands (rm -rf /, DROP DATABASE, etc.).
Override: MATCHA_SHIELD_OFF=true
Full ruleset: AGENTS.md (quick reference) → skills/matcha/SKILL.md (complete)
Available agents: @matcha-planner, @matcha-finder, @matcha-auditor, @matcha-reviewer, @matcha-cleaner, @matcha-debugger
Commands: /matcha:status, /matcha:why, /matcha:audit, /matcha:review, /matcha:intensity
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.
- 3d ago First seen · 36 lines · 537 tokens per session scan C d1c3faa06f30
matcha CLAUDE.md is an instructions file published in the GitHub repository plumpslabs/matcha (1 stars, last pushed 24d ago), licensed MIT. It adds 537 tokens to every session, about $0.0027 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
assaio AGENTS.md
Instructions for assaio/assaio, covering agents.md — assaio, what this is, code philosophy (non-negotiable), code standards (enforced + human-reviewed) and honesty rules (product-critical).
jcodemunch-mcp CLAUDE.md
Claude Code instructions for jgravelle/jcodemunch-mcp, covering jcodemunch-mcp — project brief, current state, key files, cli subcommands and architecture notes.
jcodemunch-mcp AGENTS.md
AGENTS.md instructions for jgravelle/jcodemunch-mcp, covering code exploration policy, session-aware routing and model-driven tool tiering.
assaio CLAUDE.md
Instructions for assaio/assaio, a project described as: Is your AI coding spend delivering? Offline-first analytics for Claude Code, Codex CLI, Gemini CLI, GitHub Copilot CLI and Cline — $/100 AI lines per project, every verdict carrying its own confidence, a self-contained dashboard, exec plugins in any language, and…
agent-toolkit AGENTS.md
Instructions for Arnosdeus/agent-toolkit, covering shared agent instructions, repository inspection, existing architecture, frontend and ui quality and testing.
kluris CLAUDE.md
Instructions for ngvoicu/kluris, covering claude.md, what this is, knowledge base, build & test and architecture.