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/openpi-dev/openpi/agents-mdgit clone --depth 1 https://github.com/openpi-dev/openpiWhat 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 | $0.01387 | $0.01387 |
| Opus 5 | $0.00694 | $0.00694 |
| Sonnet 5 | $0.00277 | $0.00277 |
| Haiku 4.5 | $0.00139 | $0.00139 |
Grade A, and why
openpi 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 2d 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 — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
OpenPI agent contract
OpenPI is a Pi-native capability layer, not a second agent operating system. Extend Pi at its existing seams and preserve its lifecycle, vocabulary, and sources of truth.
Design stance
Build for model leverage
- Prefer capabilities whose usefulness increases as models improve.
- Give the model small, orthogonal mechanisms and high-fidelity feedback. Let the model decide strategy, decomposition, delegation, and synthesis.
- Encode safety and execution invariants in the runtime. Do not encode a preferred reasoning process, keyword router, or fixed orchestration workflow unless correctness requires it.
- Before adding a framework abstraction, ask whether a stronger model could use the underlying Pi primitive directly. If yes, expose or strengthen that primitive instead.
Keep ownership clear
The model owns judgment:
- understanding intent and choosing an approach;
- deciding whether and how to delegate;
- adapting to evidence and synthesizing results;
- deciding when user input is needed.
The runtime owns enforceable facts:
- permissions, trust, isolation, and capability boundaries;
- concurrency, call, time, and resource limits;
- lifecycle, cancellation, cleanup, persistence, and recovery;
- atomicity, idempotency, replay safety, and fail-closed outcomes;
- observable execution state and exact terminal evidence.
Do not rely on prompts for runtime invariants. Do not move model judgment into a rigid state machine merely because it is easier to test.
Preserve Pi-native composition
- Pi remains the source of truth for Sessions, provider/model selection, Skills, project Trust, and ordinary tools.
- Reuse Pi events, messages, tools, extension hooks, and session lifecycle before introducing parallel storage or control planes.
- Keep model-facing schemas compact and progressively disclose specialized capabilities and detailed instructions only when needed.
- Treat canonical execution facts, model-visible context, and operator-facing UI as distinct projections. Never infer completion from labels or presentation state.
- Prefer one general composable capability over several workflow-specific commands.
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.
- 2d ago First seen · 89 lines · 1,387 tokens per session scan A 66bc8e442646
openpi AGENTS.md is an instructions file published in the GitHub repository openpi-dev/openpi (100 stars, last pushed 2d ago), licensed MIT. It adds 1,387 tokens to every session, about $0.0069 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
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.