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/snapdragonpartners/maestro/claude-mdgit clone --depth 1 https://github.com/SnapdragonPartners/maestroWrote 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/snapdragonpartners/maestro/claude-md)<a href="https://agentmods.dev/instructions/snapdragonpartners/maestro/claude-md"><img src="https://agentmods.dev/badge/instructions/snapdragonpartners/maestro/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 | $0.03860 | $0.03860 |
| Opus 5 | $0.01930 | $0.01930 |
| Sonnet 5 | $0.00772 | $0.00772 |
| Haiku 4.5 | $0.00386 | $0.00386 |
Grade A, and why
maestro 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 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.
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 — 340 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Operating instructions for Claude Code in this repository. AGENTS.md is a
tracked symbolic link to this file (Git mode 120000; verify with
git ls-files -s AGENTS.md), so edit CLAUDE.md only and both entry points
stay in sync. Tooling that enumerates only regular files may not list
AGENTS.md; that does not mean it is absent.
Project And Authority
Maestro is a Go application for coordinating AI agents to turn product intent into reviewed repository changes. Active development is the breaking v2 redesign. v1 remains the current implementation until replaced, but is frozen: do not fix v1 defects unless they block v2 work.
Use the source appropriate to the question.
For current runtime behavior, precedence is:
- Code and tests.
- Canonical FSM docs:
pkg/pm/STATES.md,pkg/architect/STATES.md, andpkg/coder/STATES.md. README.md, and the routing table below for where to look.deprecatedv1 docs, as unverified orientation only.
For v2 design intent, precedence is:
- Accepted ADRs (
status = "live") indocs/adr/. - Live phase artifacts in
docs/v2/phase_x/. - The roadmap and other cross-phase documents in
docs/v2/. - Historical ADRs 0001–0016, as non-binding context only.
Documents with status = "archive" carry no authority. When code and a v2 ADR
differ, the code describes today and the ADR describes the accepted direction.
Start with docs/README.md, docs/adr/README.md, and the applicable phase
README.md; do not treat a referenced deprecated document as authoritative.
Binding V2 Workflow
docs/v2/process_build.md is the binding working agreement for building v2.
It binds Claude, Codex, and DR and wins over this file if they disagree. The
current phase scope and plan at docs/v2/phase_x/plan_scope.md binds work in
that phase. docs/v2/plan_roadmap.md defines the phase sequence and
cross-phase decisions.
Throughout this file and the v2 docs, phase_x is a placeholder for the
current phase number — the real directories are docs/v2/phase_0/,
phase_1/, phase_2/, and so on. The same convention appears in ADR 0017
and process_build.md, so it is written this way here to match rather than
diverge.
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 · 340 lines · 3,860 tokens per session scan A 0f0ceaf2d460
maestro CLAUDE.md is an instructions file published in the GitHub repository SnapdragonPartners/maestro (24 stars, last pushed today), licensed MIT. It adds 3,860 tokens to every session, about $0.0193 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
codeg AGENTS.md
AGENTS.md instructions for xintaofei/codeg, covering agents.md, 项目概述, 技术栈, 代码检查与测试（任务完成后进行必要的检查） and 前端.
claude-context CLAUDE.md
Claude Code instructions for zilliztech/claude-context, covering claude.md, monorepo layout, commands, tests and running the mcp server locally.
openase AGENTS.md
Instructions for PacificStudio/openase, covering workspace notes and environment.
mue-x CLAUDE.md
Instructions for KorroAi/mue-x, covering mue — self-evolving ai agent for claude code, what you are in mue mode, commands, mode behavior and auto-start.
chiplab AGENTS.md
Instructions for veecle/chiplab, covering agents.md, what is chiplab, 1. install / connect chiplab, 2. verify the connection and 3. pick the example for the board you want.
codeg CLAUDE.md
Claude Code instructions for xintaofei/codeg, covering claude.md, 项目概述, 技术栈, 代码检查与测试（任务完成后进行必要的检查） and 前端.