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/spytensor/openmozi/agents-mdgit clone --depth 1 https://github.com/spytensor/openmoziWhat 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 | $0.02236 | $0.02236 |
| Opus 5 | $0.01118 | $0.01118 |
| Sonnet 5 | $0.00447 | $0.00447 |
| Haiku 4.5 | $0.00224 | $0.00224 |
Grade A, and why
openmozi 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 — 146 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Repository Instructions for Coding Agents
This file applies to Codex, Claude Code, Gemini CLI, and any other coding agent working in this repository.
The repo-level source of truth is docs/CONSTITUTION.md. Follow it for all architecture, delegation, and release decisions.
Non-Negotiable Rules
- Prompt text is not the execution engine. Runtime owns worker launch, sandboxing, health, state, fallback, and verification.
- Never invent worker progress, fake
queued/completedstates, or hide a deterministic runtime failure behind vague generic apologies. - Only claim capabilities that are actually registered and currently ready in this runtime.
- Treat user-defined workspace skills and workspace agents as first-class extensions; they must follow the same managed-worker contract as built-in flows.
- Do not ship delegation changes unless a real complex task has been driven end-to-end on the build and the evidence recorded. The automatic gate was removed (see
docs/CONSTITUTION.md§14) — nothing checks this for you. - Do not use generic shell execution as the transport for Claude Code, Codex, Gemini, or other external AI CLIs; those belong to managed-worker skills/adapters.
- Do not treat a feature change as complete until relevant local tests have been run and passed; if new behavior lacks coverage, add or update tests in the same change.
- Treat every bug report or feedback item as investigation-only until the user reviews the evidence and explicitly approves an implementation option. Follow the decision gate in
docs/CONSTITUTION.md§6. - Load
skills/mozi-development/SKILL.mdbefore adding or changing anything in this codebase. It carries the existing seams to reuse and the wiring proof a change needs; a raw code scan misleads you here (dead exports, docs describing intent rather than behaviour, a second implementation winning silently) and unit tests do not catch it. - Never publish to the open-source repository freehand. Load and follow
skills/public-release/SKILL.mdevery time — it carries the review gates the export script cannot enforce. Nothing that only makes sense inside this tree may appear in the public one: not in files, not in a commit message, not in a tag annotation, not in a release note. This has already gone wrong once (the tooling's own commit message published this project's name, version and commit sha on every sync) and cost a rewrite of published history.
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 · 146 lines · 2,236 tokens per session scan A 1ec20f149775
openmozi AGENTS.md is an instructions file published in the GitHub repository spytensor/openmozi (192 stars, last pushed 25d ago), licensed MIT. It adds 2,236 tokens to every session, about $0.0112 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
lobu AGENTS.md
Instructions for lobu-ai/lobu, covering repo map, unrecoverable — never do these, facts you cannot derive, ship a change and how to work.
Memoh AGENTS.md
AGENTS.md instructions for felinics/Memoh, covering agents.md, project overview, architecture overview, tech stack and backend (go).
lobu CLAUDE.md
Instructions for lobu-ai/lobu, a project described as: Open-source control plane and runtime for organisational agents: shared company context, isolated execution, approvals and MCP.
Memoh CLAUDE.md
Claude Code instructions for felinics/Memoh, a project described as: ✨ The open-source multi-agent platform. Every agent gets its own computer, desktop, network, and long-term memory. You can bring your own key, or host your coding agent like Claude Code, Codex and so on.
InvestSkill GEMINI.md
Instructions for yennanliu/InvestSkill, covering investskill — gemini cli setup & usage guide, installation & setup, quick start, navigate to the investskill directory and start gemini cli (loads gemini.md automatically).
she-love-me CLAUDE.md
Instructions for 863401402/she-love-me, covering claude.md and 唯一工作流.