mycelium AGENTS.md

Repository-specific instructions for working on Mycelium, a system for AI-assisted product development.

In plain words
What is it for?
Use them when an agent is working in the Mycelium repository and needs its recommended commands and operating rules.
Why use it?
They explain how agents should operate in that repository, including setup and available workflows.

Instructions file for CodexOpenCode

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/haabe/mycelium/agents-md
Clone the repo
git clone --depth 1 https://github.com/haabe/mycelium

Made for: Codex, OpenCode.

Per session 2,378 This file is loaded in full into every session.
When invoked 2,378 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 $0.02378 $0.02378
Opus 5 $0.01189 $0.01189
Sonnet 5 $0.00476 $0.00476
Haiku 4.5 $0.00238 $0.00238

Measured 2d ago against content hash 472d8d029670, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

mycelium AGENTS.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.

AGENTS.md · 100 lines

How it starts

The opening of the file, as written. The whole thing — 100 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agents

This repository uses Mycelium, a theory-guided harness for AI-assisted product development. See README.md for what it is and why.

If you are operating in this repo

Claude Code agents (plugin install — recommended as of v0.20.0):

/plugin marketplace add haabe/mycelium
/plugin install mycelium@haabe-mycelium

After install, run /mycelium:start (the recommended first-run command — composes setup + 10-minute discovery interview into one invocation). Skills are namespaced as /mycelium:<name> (e.g., /mycelium:interview, /mycelium:diamond-assess). Hooks (PreToolUse / PostToolUse / Stop / SessionStart / PostToolUseFailure) auto-fire. Plugin lives in ~/.claude/plugins/cache/... — your project root stays user-owned.

Invoking namespaced skills. Anthropic's plugin convention requires /<plugin>:<skill>. Two ergonomics make the prefix tax cheap:

  • Tab completion in Claude Code: /myc<Tab> expands to /mycelium:, then a few letters + <Tab> finishes the skill name.
  • Natural-language invocation: prose like "run mycelium start" or "have mycelium assess where we are" routes to the right skill. Use whichever form is faster for you.

Claude Code agents (legacy npx degit install — pre-v0.20.0): your operating manual is CLAUDE.md. Read it first. Skills invoke as /<name> without namespace. Mycelium's full enforcement layer (hooks, gates, reflexion loops, framework-guard, secret detection) is Claude-Code-specific. Both install paths supported during transition; plugin form is the recommended path.

Other agents (Codex, Cursor, Aider, Copilot, etc.): active enforcement is not yet portable. You can still read and contribute to the canvas — see "Minimal path" below. The portable surface is canvas reading + writing + decision logging; Claude-Code-specific is everything that fires automatically (PreToolUse hooks, PostToolUseFailure reflexion, secret-detection, framework-guard). For non-Claude-Code agents, the canonical instructions surface is THIS file (AGENTS.md) — Codex / Cursor / Aider / Copilot read it natively. Claude Code falls back to AGENTS.md when no CLAUDE.md exists. The framework canonical content lives at plugins/mycelium/ in the upstream haabe/mycelium repo (skill SKILL.md files are pure markdown — readable by any agent).

Read the full file on GitHub · 100 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 · 100 lines · 2,378 tokens per session scan A 472d8d029670

Subscribe to this mod's changes

mycelium AGENTS.md is an instructions file published in the GitHub repository haabe/mycelium (45 stars, last pushed 2d ago), licensed MIT. It adds 2,378 tokens to every session, about $0.0119 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.