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/agentproto/ts/agents-mdgit clone --depth 1 https://github.com/agentproto/tsWhat 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 | $0.03502 | $0.03502 |
| Opus 5 | $0.01751 | $0.01751 |
| Sonnet 5 | $0.00700 | $0.00700 |
| Haiku 4.5 | $0.00350 | $0.00350 |
Grade A, and why
ts 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 — 229 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent instructions for this repo
This file is the committed definition of done for any agent session working
in agentproto/ts — local or delegated, human-prompted or supervisor-run.
If you're an agent reading this from a prompt instead of loading it, stop:
load this file first, it overrides ad-hoc instructions.
Not to be confused with: an AIP-42 agent manifest (also named AGENT.md,
singular, loaded from <workspace>/.agents/<id>/AGENT.md —
packages/agent/src/load-agent.ts:11), or .github/AGENT.md, the cloud
reviewer/fixer harness that runs on every PR — see that file for what it
does; this file is about what you (the session driving the work) must do
before and around it.
Reach: this is a convention, not enforcement. It binds local and
delegated coding sessions that load it — nothing in the daemon reads it, and
the cloud reviewer doesn't either (it runs on .github/AGENT.md + the
aip-conventions skill from .github/agentic-review.json). Loading this
file is still on the session; it doesn't reach in and gate anything by
itself.
Definition of done for agent sessions
Done = a green local gate + an open PR. That is the terminal state for an
agent session — the handoff, not a request for permission. Your rung ends
there; review, risk-judgement, escalation, and merge are the CI plane's job,
already declared in .github/agentic-review.json + .github/workflows/ci.yml.
Each rung has exactly one owner — don't reach into the next one.
-
Green gate. Run this repo's own gate before calling anything done:
agentproto.jsondeclaresscripts.testaspnpm test(agentproto.json:7, itselfpnpm -r --filter "./packages/**" --filter "./adapters/**" test,package.json:19). Don't invent your own bar.Read the gate's exit code, not its output.
pnpm test | tail -30reportstail's exit status, not the test run's — a pipeline exits with its LAST command's status, so a red gate reads as exit 0 and you will report a passing gate that failed.${PIPESTATUS[0]}does not rescue it either: that's bash, and this repo's sessions run zsh, where it's$pipestatus[1]and the bash spelling silently expands to empty. Redirect and check$?directly:
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 · 229 lines · 3,502 tokens per session scan A 76d8aaca2f8d
ts AGENTS.md is an instructions file published in the GitHub repository agentproto/ts (5 stars, last pushed yesterday), licensed Apache-2.0. It adds 3,502 tokens to every session, about $0.0175 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
agent-plugins copilot-instructions.md
Instructions for awslabs/agent-plugins, covering project overview, technology stack, code organization, development practices and plugin development guidelines.
wanaku CLAUDE.md
Instructions for wanaku-ai/wanaku, covering wanaku, core guidelines, quick start, build info and architecture.
autogen copilot-instructions.md
Copilot instructions for microsoft/autogen, covering autogen multi-agent ai framework, working effectively, prerequisites and environment setup, python development workflow and quick validation (under 1 second each).
hatch3r CLAUDE.md
Instructions for hatch3r/hatch3r, covering hatch3r — development instructions, architecture, development commands, two-axis pillar framework (2.0.0) and orchestrator self-discipline (bypass protection).
cybara CLAUDE.md
Instructions for metaspartan/cybara, covering ai coding rules for cybara agent platform, no code comments (strict), runtime environment, import rules and ✅ do.
cybara AGENTS.md
Instructions for metaspartan/cybara, covering cybara agent rules, package manager policy (strict), command translation, typescript type safety and no comments.