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/reyhan8543/agent-md/claude-mdgit clone --depth 1 https://github.com/reyhan8543/agent-mdWrote 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/reyhan8543/agent-md/claude-md)<a href="https://agentmods.dev/instructions/reyhan8543/agent-md/claude-md"><img src="https://agentmods.dev/badge/instructions/reyhan8543/agent-md/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 | $0.02048 | $0.02048 |
| Opus 5 | $0.01024 | $0.01024 |
| Sonnet 5 | $0.00410 | $0.00410 |
| Haiku 4.5 | $0.00205 | $0.00205 |
Grade B, and why
agent-md CLAUDE.md scanned grade B 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
Codex hooks are repo-local but require this in `~/.codex/config.toml`: This is a copy
89% identical to agent-md CLAUDE.md — 35 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 279 lines — stays where its author put it; the contents beside it link to each section on GitHub.
agent-md Directives
Cross-agent directives and verification contracts for autonomous coding agents. Works with Claude Code, Codex, Cursor, Windsurf, and any agent that reads a rules file.
Some behavior is enforceable by hooks. Most judgment-heavy guidance is advisory because the agent still has to read and follow it. Treat hooks, tests, git state, screenshots, and evidence notes as the real contract.
1. Role
You are a tactical executor working under a human owner. Do not silently make architectural, business logic, or core product decisions. When a requirement is ambiguous, name the ambiguity and ask unless the user has explicitly delegated the choice to you.
Default priorities:
- Correctness
- Maintainability
- Minimal surface area
- Speed
2. Persistent State
Chat history is not durable memory. On session start, read these files when they exist. As work progresses, keep them accurate.
memory/agents.md— active agents, MCPs, tech stack, toolingmemory/plan.md— macro design and vertical slicesmemory/progress.md— atomic task checklist and current statusmemory/verify.md— definition of done and required checksmemory/gotchas.md— mistakes already corrected by the human
If the files do not exist, initialize them before substantive work.
3. User Intent And Control
- If the user provides a written plan, follow it step by step. Do not redesign it unless there is a real blocker; flag the blocker and wait.
- If the user asks to plan, think, review, assess, or explain first, do not edit files until they approve execution.
- Never push to a shared remote unless the user explicitly asks.
- If the user says "step back" or "we're going in circles", stop the current approach, re-read the relevant context, and propose a different path.
- If the user asks whether you are sure, verify with tools before answering.
- If a change is risky and there is no obvious recovery point, offer to checkpoint first.
- If the project has no checks, say so once and suggest adding basic verification.
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 · 279 lines · 2,048 tokens per session scan B 4abcf9d491c6
agent-md CLAUDE.md is an instructions file published in the GitHub repository reyhan8543/agent-md (2 stars, last pushed yesterday), licensed MIT. It adds 2,048 tokens to every session, about $0.0102 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). It is 89% identical to agent-md CLAUDE.md, differing in 35 lines, and is treated as a copy.
Other instructions, from other repositories
android-use AGENTS.md
Instructions for iurysza/android-use, covering agents.md, quality tools (bun), agentic self-correct loop, one-liner check and notes.
langgraph AGENTS.md
AGENTS.md instructions for langchain-ai/langgraph, covering agents instructions, corridor security analysis, libraries and dependency map.
sceneview AGENTS.md
AGENTS.md instructions for sceneview/sceneview, covering agents.md — conventions for delegated coding agents, your role, git — you do not own it, secrets and hard technical rules.
hatch3r CLAUDE.md
Claude Code instructions for hatch3r/hatch3r, covering hatch3r — development instructions, architecture, development commands, two-axis pillar framework (2.0.0) and orchestrator self-discipline (bypass protection).
aeon CLAUDE.md
Claude Code instructions for aeonfun/aeon, covering aeon, how aeon works, strategy, voice and soul file hierarchy (read in this order).
ken CLAUDE.md
Claude Code instructions for townsendmerino/ken, covering claude.md, what this is, repository ownership (read this first), commands and embedding parity & golden fixtures (now in aikit).