openmozi CLAUDE.md

Project instructions for OpenMozi, including its design principles and rules for using skills and recovering from problems.

In plain words
What is it for?
Use them when planning or implementing changes in the OpenMozi repository.
Why use it?
They give the coding agent the project's priorities and constraints before it changes code.

Instructions file

Install

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.

agentmods
npx agentmods add instructions/spytensor/openmozi/claude-md
Clone the repo
git clone --depth 1 https://github.com/spytensor/openmozi
Per session 5,975 This file is loaded in full into every session.
When invoked 5,975 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.05975 $0.05975
Opus 5 $0.02988 $0.02988
Sonnet 5 $0.01195 $0.01195
Haiku 4.5 $0.00598 $0.00598

Measured 2d ago against content hash 9239965da4bf, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

openmozi CLAUDE.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.

CLAUDE.md · 424 lines

How it starts

The opening of the file, as written. The whole thing — 424 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md — Instructions for Claude Code

Repo-wide constitutional rules live in docs/CONSTITUTION.md. If this file and the constitution diverge, follow the constitution and update this file.

Design Philosophy — Read This First

MOZI = Jarvis. The sole purpose of MOZI is to solve the user's problems. Every design decision must serve this goal.

  • MOZI is a personal Agent OS first. Optimize for one operator getting real work done on a real machine. Multi-tenant or platform-scale concerns must not override the current product goal, but interfaces should remain extensible so future growth does not require a rewrite.
  • The chosen LLM is replaceable. Models are schedulers/reasoners that can be swapped. The runtime is the durable operating substrate.
  • Skills are part of execution, not decoration. Prefer Anthropic-compatible SKILL.md style assets and local skill packs over bespoke one-off prompt logic when the same workflow can be expressed as a reusable skill.
  • Default to autonomous recovery. If the task can be recovered safely by retrying, repairing, or choosing a better route, do it. Interrupt the user only for approval, ambiguity, irreversible risk, or a hard block.
  • The LLM Brain makes ALL decisions. Infrastructure (gateway, TEL, tools, proactive engine) exists only to execute what the Brain decides. Never add hardcoded logic that overrides, second-guesses, or teaches the LLM how to think — that's the Brain's job.
  • If a change doesn't help the user get things done faster or better, don't make it. "More autonomous", "more observable", "more intelligent infrastructure" are not valid justifications. The user doesn't care about internal metrics, system status reports, or self-monitoring — they care about results.
  • Don't build scaffolding around the Brain. No fake user messages, no invented worker progress, no placeholder "queued" success, and no vague fallback narration that hides runtime truth.
  • Complexity must be justified by user impact. Every line of code is a liability. Before adding a feature, ask: "Will the user notice this?" If not, don't ship it.

Read the full file on GitHub · 424 lines

Changes

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.

  1. 2d ago First seen · 424 lines · 5,975 tokens per session scan A 9239965da4bf

Subscribe to this mod's changes

openmozi CLAUDE.md is an instructions file published in the GitHub repository spytensor/openmozi (210 stars, last pushed 26d ago), licensed MIT. It adds 5,975 tokens to every session, about $0.0299 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.