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/permit0-ai/permit0/claude-mdgit clone --depth 1 https://github.com/permit0-ai/permit0Wrote 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/permit0-ai/permit0/claude-md)<a href="https://agentmods.dev/instructions/permit0-ai/permit0/claude-md"><img src="https://agentmods.dev/badge/instructions/permit0-ai/permit0/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 | $0.01759 | $0.01759 |
| Opus 5 | $0.00879 | $0.00879 |
| Sonnet 5 | $0.00352 | $0.00352 |
| Haiku 4.5 | $0.00176 | $0.00176 |
Grade A, and why
permit0 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 5d 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 — 144 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Commands
# Build
cargo build --workspace
cargo build --release --workspace
# Test (CI uses nextest)
cargo nextest run --workspace
cargo test --workspace
# Run a single test
cargo nextest run -p permit0-engine <test_name>
cargo test -p permit0-scoring -- scorer::tests::more_flags_higher_score
# Lint (warnings are errors)
cargo clippy --workspace --all-targets -- -D warnings
# Format
cargo fmt --all
cargo fmt --all --check
# Supply-chain audit
cargo deny check
# Calibration tests (run after changing packs or profiles)
cargo run -- calibrate test
cargo run -- calibrate test --corpus corpora/calibration --profile fintech
# Pack testing
cargo run -- pack test packs/stripe
cargo run -- pack validate packs/stripe
# Start admin dashboard
cargo run -- serve --ui --port 9090
RUSTFLAGS="-D warnings" is set in CI; all warnings are fatal.
Python bindings (permit0-py)
The Python bindings crate is excluded from default-members, so cargo build (no flags) skips it. Two ways to build it:
# Canonical: build + install editable wheel into the active Python env
cd crates/permit0-py
maturin develop # uses whichever Python `maturin` resolves to
# Plain cargo (also works thanks to .cargo/config.toml on macOS)
cargo build --workspace # builds every crate including permit0-py
cargo build -p permit0-py # builds permit0-py only
macOS notes:
.cargo/config.tomladds-undefined dynamic_lookupforaarch64-apple-darwinandx86_64-apple-darwinso PyO3'sextension-modulecdylib links without a libpython.- If both a conda env and a
.venvare active, maturin refuses with a dual-env error —unset VIRTUAL_ENV(orunset CONDA_PREFIX) to disambiguate.
Architecture
permit0 is a deterministic, rule-based permission framework for AI agent tool calls. There is no LLM in the hot path — the decision pipeline is pure Rust.
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.
- 5d ago First seen · 144 lines · 1,759 tokens per session scan A e4fc07676e6e
permit0 CLAUDE.md is an instructions file published in the GitHub repository permit0-ai/permit0 (185 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 1,759 tokens to every session, about $0.0088 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 2026-08-30.
Other instructions, from other repositories
getregula AGENTS.md
AGENTS.md instructions for kuzivaai/getregula, covering regula — agent instructions, build & test, verify (run all before claiming done), key constraints and quality checkpoints.
ClawCrate AGENTS.md
Instructions for manuelpenazuniga/ClawCrate, covering agents.md — clawcrate context for codex, primary sources, project summary, non-negotiable architecture rules and engineering guardrails.
h5i AGENTS.md
AGENTS.md instructions for h5i-dev/h5i, a project described as: Secure, auditable browser for AI agents. Pure Rust, no Chromium or V8. 3× faster and 86% less peak memory than headless Chromium.
ClawCrate CLAUDE.md
Instructions for manuelpenazuniga/ClawCrate, covering claude.md — clawcrate development guide, project overview, core architecture, key design decisions — do not deviate and tech stack.
cli AGENTS.md
AGENTS.md instructions for googleworkspace/cli, covering agents.md, project overview, build & test, changesets and architecture.
turso AGENTS.md
AGENTS.md instructions for tursodatabase/turso, covering turso agent guidelines, quick reference, testing, running tests and test organization.