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/selfradiance/agentgate/agents-mdgit clone --depth 1 https://github.com/selfradiance/agentgateWhat 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 | $0.01113 | $0.01113 |
| Opus 5 | $0.00557 | $0.00557 |
| Sonnet 5 | $0.00223 | $0.00223 |
| Haiku 4.5 | $0.00111 | $0.00111 |
Grade A, and why
agentgate 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 yesterday.
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 — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Commands
npm installinstalls project dependenciesnpm run devstarts the local Fastify server with reloadnpm run buildcompiles TypeScriptnpm run testruns the automated test suitenpm run lintruns the TypeScript typecheck used as linting
Conventions
- Keep the implementation minimal and local-first
- Use Fastify route handlers with zod parsing for request validation
- Use SQLite through
better-sqlite3; prefer:memory:in tests - Keep bond settlement logic in
src/service.ts - Add tests for state transitions through Fastify injection rather than only unit-level helpers
AgentGate: Instructions for Any AI Assistant
This repository is developed with help from AI. These rules apply regardless of model or tool.
User skill level
- The user is a beginner.
- Use baby steps.
- Provide clear, small, actionable steps.
Non-negotiable workflow (always)
For any meaningful change:
- Make the smallest safe change.
- Verify:
- run typecheck/build (as applicable)
- run demo script(s) / minimal tests
- Commit with a clear message.
- Push to GitHub immediately after committing. Do not leave commits unpushed unless explicitly told to hold off.
- Update README with:
- what changed
- how to verify
- next steps
If verification fails, stop and fix before proceeding.
Security priorities
- Treat auth/token/nonce/replay/transport logic as high risk.
- Prefer safety and clarity over speed.
- Never weaken security checks to “make it work”.
- Log security-relevant events (without leaking secrets).
Secrets and sensitive data
- Never hardcode secrets.
- Never commit tokens/keys/.env.
- Use example files and documented env vars instead.
Working style
- Ask before large refactors.
- Prefer small diffs and incremental steps.
- Keep diffs under ~100 lines per change. If a change exceeds 300 lines, stop and break it into smaller pieces before proceeding.
- When uncertain, propose 2 options with tradeoffs and default to safer.
- Do not claim something is verified unless you list what you ran/checked.
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.
- yesterday First seen · 116 lines · 1,113 tokens per session scan A 0a32248256c4
agentgate AGENTS.md is an instructions file published in the GitHub repository selfradiance/agentgate (5 stars, last pushed 4mo ago), licensed MIT. It adds 1,113 tokens to every session, about $0.0056 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
nono AGENTS.md
AGENTS.md instructions for nolabs-ai/nono, covering nono - development guide, project overview, library vs cli boundary, build & test and build everything.
nono CLAUDE.md
Claude Code instructions for nolabs-ai/nono, a project described as: secure multiplexed execution paths for agents - zero trust, zero setup, zero latency.
arcjet-js AGENTS.md
Instructions for arcjet/arcjet-js, covering agent guidance, examples live in arcjet/examples and integration work: review before a pr.
prismor CLAUDE.md
Instructions for PrismorSec/prismor, covering prismor security — claude.md, prismor runtime protection, cloaking (secret prevention) and working in this repo.
ave CLAUDE.md
Instructions for aveproject/ave, covering claude.md — aveproject/ave, project, the critical distinction — read this twice, record schema v1.1.0 and adding a record.
Agent-Security-Regression-Harness AGENTS.md
Instructions for OWASP/Agent-Security-Regression-Harness, covering ai agent guide: owasp agent security regression harness, project overview, architecture, contribution workflow and what needs doing.