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/paiml/paiml-mcp-agent-toolkit/agents-mdgit clone --depth 1 https://github.com/paiml/paiml-mcp-agent-toolkitWrote 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/paiml/paiml-mcp-agent-toolkit/agents-md)<a href="https://agentmods.dev/instructions/paiml/paiml-mcp-agent-toolkit/agents-md"><img src="https://agentmods.dev/badge/instructions/paiml/paiml-mcp-agent-toolkit/agents-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 5.1 | $0.00599 | $0.00599 |
| Opus 5 | $0.00300 | $0.00300 |
| Sonnet 5 | $0.00120 | $0.00120 |
| Haiku 4.5 | $0.00060 | $0.00060 |
Grade A, and why
paiml-mcp-agent-toolkit AGENTS.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 6d 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 — 46 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Antigravity Global Rules for PMAT
Welcome to the paiml-mcp-agent-toolkit (PMAT) codebase! PMAT is a zero-config AI context generation and code quality toolkit.
Agent Guidelines
- Use the PMAT CLI via the configured MCP Server for quality checks and analysis.
- Follow the rules defined in
.agents/rules/for all operations. - Maintain test coverage above 95% (as detailed in the rules).
- Adhere to the autonomous verify loop protocol for self-correction.
When making code changes, ensure that PMAT quality gates (via the pmat-quality-feedback.sh hook) pass. If it fails, fix the code according to the feedback!
Contract-first, and how it is enforced
PMAT is contract-first: a rule's behaviour is declared in contracts/*.yaml and
bound to the code by #[provable_contracts_macros::contract(...)], so the
contract is checked by the build rather than by review. Write or update the
contract before the code it governs.
Run pmat verify before every commit. It runs the gate set CI enforces —
format, complexity, satd, clippy, tests — fail-fast, so green here means green
in CI. The pre-commit hook and pmat quality-gate both skip clippy and tests;
pmat verify does not.
pmat comply check audits the 157 registered rules. Two of them are worth
knowing before you touch a number in a config file:
- CB-2101 (threshold coherence) — every scalar threshold in
.pmat-metrics.tomlmust bind to a metric in.pmat-ratchet.toml. A limit that is breached at HEAD, or so far from the measurement that nothing could ever reach it, fails. An unbound threshold fails too: a number enforced by nothing reads as enforcement and enforces nothing. - CB-2102 (ratchet) — baselines may only improve. The gate does not read the
number; it re-runs the
commandrecorded beside it. A metric that cannot be measured FAILS, because "we could not measure it" must never be recorded as "it did not regress".
That last rule generalises, and it is the one to internalise: a measurement with no failing condition is indistinguishable from a correct one. Prefer the compiler over a text search whenever the question is "does this construct appear in shipped code" — pmat is a linter, so its own source contains as data every pattern it hunts, and grepping it for a defect marker returns its own detectors. Before trusting any new gate, apply the defect it targets and confirm it turns red, then restore and confirm it turns green again.
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.
- 6d ago First seen · 46 lines · 599 tokens per session scan A 5bc9fe6e5f33
paiml-mcp-agent-toolkit AGENTS.md is an instructions file published in the GitHub repository paiml/paiml-mcp-agent-toolkit (164 stars, last pushed yesterday), licensed MIT. It adds 599 tokens to every session, about $0.0030 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
wanaku CLAUDE.md
Claude Code instructions for wanaku-ai/wanaku, covering wanaku, core guidelines, quick start, build info and architecture.
oracle-forms-mcp CLAUDE.md
Claude Code instructions for aoreshkov/oracle-forms-mcp, covering claude.md, layout, invariants, gotchas and commands.
GhostDesk AGENTS.md
AGENTS.md instructions for YV17labs/GhostDesk, covering names — five levels, and none overflows into the next, ports & adapters — three module shapes, and no fourth, families — a shared prefix names a standard, never a theme, modules — two files, two jobs, never merged and configuration — one seam per config, decided…
ops-brain CLAUDE.md
Claude Code instructions for TheK3nsai/ops-brain, covering ops-brain, surface (15 tools), architecture constraints, client-scope disclosure guard and cross-client gate behavior.
geolocation-mcp CLAUDE.md
Claude Code instructions for 7mza/geolocation-mcp, covering claude.md, stack, commands, setup graalvm and run on jvm.
agentic-obs CLAUDE.md
Instructions for ironystock/agentic-obs, covering claude.md, project overview, project structure, key technologies and communication flow.