metacog: Instructions file for Claude Code

CLAUDE.md

metacog CLAUDE.md is an instructions file for Claude Code from houtini-ai/metacog. It costs 889 tokens per session, scanned A, original, Apache-2.0.

Project instructions for Metacog, a Claude Code hook plugin that watches an agent's working state and signals when its behaviour may be abnormal. Hooks are small actions that run automatically after tools or user messages.

In plain words
What is it for?
Use it when changing Metacog's cognitive-state checks, error detection, task-drift tracking, session learning, or hook behaviour.
Why use it?
It gives developers the rules needed to understand and safely modify Metacog's two hooks, seven monitoring checks, and temporary session state.

Instructions file for Claude Code

Written for Claude Code: PostToolUse hook event. Also seen: mentions subagents; mentions Claude Code.

This is houtini-ai/metacog's own configuration. It tells Claude Code how to work on metacog itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything metacog configures →

Reuse

Borrowing it

Nothing to install: this file belongs to houtini-ai/metacog. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/houtini-ai/metacog/main/CLAUDE.md
Clone the repo
git clone --depth 1 https://github.com/houtini-ai/metacog

Made for: Claude Code.

Wrote 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.

agentmods badge for metacog CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/houtini-ai/metacog/claude-md/github.svg)](https://agentmods.dev/instructions/houtini-ai/metacog/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/houtini-ai/metacog/claude-md"><img src="https://agentmods.dev/badge/instructions/houtini-ai/metacog/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for metacog CLAUDE.md

Your own site · 80×15
<a href="https://agentmods.dev/instructions/houtini-ai/metacog/claude-md"><img src="https://agentmods.dev/badge/instructions/houtini-ai/metacog/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 889 This file is loaded in full into every session.
When invoked 889 The same file — it is already loaded in full.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00889 $0.00889
Opus 5 $0.00445 $0.00445
Sonnet 5 $0.00178 $0.00178
Haiku 4.5 $0.00089 $0.00089

Measured 9d ago against content hash 6ea20a04a8ae, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

metacog 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 9d 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.

CLAUDE.md · 67 lines

How it starts

The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Metacog

Claude Code hook plugin that gives the agent real-time awareness of its own cognitive state.

Architecture

Two hooks, one state file:

  • src/hook.js (PostToolUse) - fires after every tool call, evaluates 7 senses, outputs a signal only when something is abnormal. Silent exit 0 = everything fine. JSON stdout = signal for the agent.
  • src/digest-inject.js (UserPromptSubmit) - fires on every user message. First message of session: injects behavioural digest + session retrospective. Every message: captures task fingerprint for drift detection, records user interaction metadata.
  • .claude/metacog.state.json - ephemeral rolling window of last 20 actions. Written by hook.js, read by both hooks. Reset on session change. Do not persist or version control this file.

The 7 senses (src/senses/)

Sense File What it watches
O2 o2.js Token velocity vs baseline
Chronos chronos.js Time and steps since last user interaction
Nociception nociception.js Consecutive similar errors (Layer 1 + Layer 2 escalation)
Spatial spatial.js git grep for importers after writes
Vestibular vestibular.js Repeated identical actions
Echo echo.js Write streaks without test runs, self-test pattern
Drift drift.js Recent actions diverging from task fingerprint

Key principles when modifying senses

  • Silent by default. A sense that fires too often is worse than one that never fires. No news is good news.
  • Every sense has a cooldown. After firing, suppress for N turns. One nudge is enough.
  • Graceful degradation. Any error in any sense must be caught. If the hook crashes, the agent loses proprioception but keeps working. Never let the nervous system kill the brain.
  • Inform, don't command. Layer 1 signals are observations the agent can dismiss. Only Layer 2 (nociception escalation) forces structured reflection, and it earns that right by requiring repeated failure first.
  • Subagent awareness. Agent tool calls are delegations, not grinding. Suppress O2 and Chronos on delegation returns.
  • Cooldown ticking. If you add a new cooldown field to state, you must also tick it down in state.js:appendAction().

Read the full file on GitHub · 67 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. 9d ago First seen · 67 lines · 889 tokens per session scan A 6ea20a04a8ae

Subscribe to this mod's changes

metacog CLAUDE.md is an instructions file published in the GitHub repository houtini-ai/metacog (5 stars, last pushed 4mo ago), licensed Apache-2.0. It adds 889 tokens to every session, about $0.0044 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.

Related

Other instructions, from other repositories

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).

microsoft/vscode · 5,001 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,469 tokens

deepseek-harness AGENTS.md

AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.

deepseek-ai/deepseek-harness · 3,735 tokens