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/earthwalker17/agent-os/claude-mdgit clone --depth 1 https://github.com/earthwalker17/agent-osWrote 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/earthwalker17/agent-os/claude-md)<a href="https://agentmods.dev/instructions/earthwalker17/agent-os/claude-md"><img src="https://agentmods.dev/badge/instructions/earthwalker17/agent-os/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 | $0.02447 | $0.02447 |
| Opus 5 | $0.01223 | $0.01223 |
| Sonnet 5 | $0.00489 | $0.00489 |
| Haiku 4.5 | $0.00245 | $0.00245 |
Grade C, and why
agent-os CLAUDE.md scanned grade C with 1 finding 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 5d 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
run destructive shell commands (`rm -rf`, `git push --force`, How it starts
The opening of the file, as written. The whole thing — 179 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Operating guide for Claude Code and any other coding agent working on this
repository. This file is intentionally short, stable, and phase-independent.
For the system's shape (files, pipelines, invariants) read ARCHITECTURE.md;
for the project's history and status read ROADMAP.md; for long-term direction
read BLUEPRINT.md. The public README.md is a landing page, not a status file.
1. Project Mission
Agent OS is a lightweight local-first project cockpit — a small Agent Operating System for managing multiple long-term projects through a single web chat surface. It combines project-scoped conversations, structured markdown memory, an orchestration layer, and a bounded execution layer that hands work to a Coding Agent inside a sandboxed workspace, then delivers the result through audited Git/GitHub and deployment contracts. The goal is not a general-purpose assistant or a heavyweight agent platform; it is a clear, controllable place for a single builder to plan, decide, and execute project work.
2. Core Architecture Principles
- Local-first. Filesystem + SQLite + FastAPI + React. No cloud services, no queues, no external infra dependencies for the core.
- Project isolation. Each project has its own conversations, memory files, and execution workspace. Crossings are deliberate and bounded.
- Structured markdown memory. Important state lives in readable, editable
.mdfiles, not buried in chat history. - Main agent = brain. Planner / memory steward / orchestrator. Does not
edit code under
repo/. - Coding Agent = hands. Bounded executor inside the project's
execution_workspaces/{project_id}/repo/. Does not edit project memory or other projects' workspaces. - All execution goes through
ProjectSandbox+ToolRuntime. No rawos/pathlibaccess to repo paths, no rawsubprocess. Every tool call routes throughresolve_repo_path()/resolve_under()orvalidate_command().
3. Agent Roles
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.
- 5d ago First seen · 179 lines · 2,447 tokens per session scan C 4c83410b2bc4
agent-os CLAUDE.md is an instructions file published in the GitHub repository earthwalker17/agent-os (21 stars, last pushed yesterday), licensed Apache-2.0. It adds 2,447 tokens to every session, about $0.0122 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
flow CLAUDE.md
Claude Code instructions for AixleHQ/flow, covering claude.md, documentation index, before pushing a branch or opening/updating a pr, docker is required and test parallelization.
psyche-build AGENTS.md
AGENTS.md instructions for OpenCoven/psyche-build, covering psyche build agent guide, repository role, planning and evidence contract, canonical routing and deterministic entrypoints.
seektty AGENTS.md
AGENTS.md instructions for Hilbert-beinghappy/seektty: This repository ships one out-of-tree DeepSeek Harness Bundle. Harness remains the only owner of Agent, Session, model, settings, permissions, Profile, plugin, and persistence state.
open-bridge AGENTS.md
AGENTS.md instructions for bks-lab/open-bridge, covering the bridge — agent instructions, required reading, session start detection (automatic), theme and agents.
agents CLAUDE.md
Instructions for eloylp/agents, covering claude.md, project overview, directory structure, config model and build & run.
agent-skills AGENTS.md
AGENTS.md instructions for simota/agent-skills, covering agents.md — claude-skills, repository purpose, language policy, repository structure and skill authoring conventions.