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/ljchang/mecha/claude-mdgit clone --depth 1 https://github.com/ljchang/mechaWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/ljchang/mecha/claude-md)<a href="https://agentmods.dev/instructions/ljchang/mecha/claude-md"><img src="https://agentmods.dev/badge/instructions/ljchang/mecha/claude-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.06754 | $0.06754 |
| Opus 5 | $0.03377 | $0.03377 |
| Sonnet 5 | $0.01351 | $0.01351 |
| Haiku 4.5 | $0.00675 | $0.00675 |
Grade A, and why
mecha 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 today.
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 — 472 lines — stays where its author put it; the contents beside it link to each section on GitHub.
mecha — CLAUDE.md
What this is
An agent harness whose purpose is to make a local open-weight model into a usable personal assistant: wired into enough personal context to be worth asking (mail, calendar, a knowledge graph), able to reach the world through a reviewed surface, and safe to hold all of that at once. Almost every design decision here is downstream of that last clause — a personal assistant has private data, third-party content and a way to send by definition, so the lethal trifecta is the permanent condition rather than an edge case.
Four crates:
mecha-core/— the library. Knows nothing about any CLI or application.mecha-cli/— themechabinary. Thin; all logic belongs in core.mecha-mail/— mail, calendar and documents. Mail and calendar sit behind one provider-neutral MCP surface;mecha-docsis a fourth binary on the same library, sharing its OAuth and token lifecycle.mecha-slack/— the Slack transport. Nomecha-coredependency, ever; that is the whole reason it is a crate, and it is checkable inCargo.toml.
Interfaces: mecha run (one-shot), mecha chat (readline REPL), mecha tui
(full-screen; the input line stays live so you can steer a run in flight), and
mecha batch / mecha eval for fan-out.
The user-facing half of this reasoning is published at
https://docs.mecha-factory.ai/; website/docs/principles.md is that site's
restatement of the invariants below, without the incident behind each one.
Where everything else lives. This file holds only what any session may
need on any run; it is the expensive file, because it rides in every agent's
context. docs/ARCHITECTURE.md holds the per-subsystem invariants —
read its section before changing a subsystem it names, and put a new
subsystem's invariants there, not here. docs/README.md is the map of all
the rest — which document holds what, and the routing question to ask
before writing a line anywhere. Two of its rules govern this file: state the
rule, then the incident in one sentence, never the other way round; and where
a topic doc covers something, point at it rather than restating it, because
duplication is how documents drift into disagreeing.
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.
- today Changed · +12 lines · +180 tokens per session 2520350dd026
- yesterday Changed · +39 lines · +612 tokens per session 1e35c6007c11
- 5d ago First seen · 421 lines · 5,962 tokens per session scan A 464dee482536
mecha CLAUDE.md is an instructions file published in the GitHub repository ljchang/mecha (6 stars, last pushed today), licensed MIT. It adds 6,754 tokens to every session, about $0.0338 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
eigent AGENTS.md
AGENTS.md instructions for eigent-ai/eigent, covering eigent agent instructions, ui work, product terminology and validation.
Agentlas-OS GEMINI.md
Gemini CLI instructions for agentlas-ai/Agentlas-OS, covering gemini cli adapter, public release boundary, startup, default behavior and global command.
autopus-adk AGENTS.md
AGENTS.md instructions for Insajin/autopus-adk, covering autopus-adk harness, installed components, language policy, autopus branding and document storage.
HARTOS CLAUDE.md
Claude Code instructions for hertz-ai/HARTOS, covering claude.md, project overview, master key - ai exclusion zone, branch discipline — main branch only (mandatory) and sibling repos — canonical filesystem paths.
orloj AGENTS.md
AGENTS.md instructions for OrlojHQ/orloj, covering orloj agent instructions, must-follow sync rules and working style.
autopus-adk CLAUDE.md
Claude Code instructions for Insajin/autopus-adk, covering autopus-adk harness, 설치된 구성 요소, rule isolation, language policy and core guidelines.