thoughtbox CLAUDE.md

thoughtbox CLAUDE.md is an instructions file for coding agents from Kastalien-Research/thoughtbox. It costs 1,297 tokens per session, scanned B, original, MIT.

A Claude Code instruction file for the Thoughtbox project. It requires verification before making claims and defines the repository’s commit-message format.

In plain words
What is it for?
Use it when Claude Code works in Thoughtbox and needs rules for checking files, running tests, and writing commit messages.
Why use it?
It helps prevent unverified statements about files or tests and keeps Git commits consistent with the project’s conventions.

Instructions file

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/kastalien-research/thoughtbox/claude-md
Clone the repo
git clone --depth 1 https://github.com/Kastalien-Research/thoughtbox

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 thoughtbox CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/kastalien-research/thoughtbox/claude-md.svg)](https://agentmods.dev/instructions/kastalien-research/thoughtbox/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/kastalien-research/thoughtbox/claude-md"><img src="https://agentmods.dev/badge/instructions/kastalien-research/thoughtbox/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,297 This file is loaded in full into every session.
When invoked 1,297 The same file — it is already loaded in full.
Security scan B 1 finding. 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 $0.01297 $0.01297
Opus 5 $0.00648 $0.00648
Sonnet 5 $0.00259 $0.00259
Haiku 4.5 $0.00130 $0.00130

Measured yesterday against content hash cc55b2dcd345, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

thoughtbox 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 yesterday.

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.

Hub coordination (workspaces, problems, proposals, channels) is exposed over MCP as `tb.hub.*` inside `thoughtbox_execute` (SPEC-V1-INITIATIVE Phase 4.1). Call `tb.hub.register({ name })` or `tb.hub.quickJoin({ name, wor
CLAUDE.md · 133 lines

How it starts

The opening of the file, as written. The whole thing — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Claude Code Instructions

@AGENTS.md


CRITICAL: Always Verify

NEVER say "probably", "maybe", "might be", "could be", or other conditional hedging.

Before making ANY claim about:

  • Whether a file exists
  • Whether code is present
  • Whether a feature works
  • Whether something was done

You MUST verify it first using the appropriate tool (Read, Glob, Grep, Bash, etc.).

Do NOT:

  • Assume something exists without checking
  • Claim something works without testing
  • Say "it should be there" without reading the file
  • Trust previous agents' claims without verification

DO:

  • Read files before claiming they contain something
  • Run tests before claiming they pass
  • Check directory contents before claiming files exist
  • Verify configurations before claiming they're correct

Commit Message Format (REQUIRED)

This project uses Conventional Commits

When creating commits, you MUST use this format:

<type>[optional scope]: <description>

[optional body]

Quick Reference

Types (use these exactly):

  • feat: New feature → CHANGELOG Added section
  • fix: Bug fix → CHANGELOG Fixed section
  • perf: Performance improvement → CHANGELOG Changed section
  • refactor: Code refactor → CHANGELOG Changed section
  • docs: Documentation only → Omitted from CHANGELOG
  • test: Tests only → Omitted from CHANGELOG
  • chore: Maintenance/tooling → Omitted from CHANGELOG
  • security: Security fix → CHANGELOG Security section
  • breaking or add !: Breaking change → MAJOR version bump

Examples

# New feature
feat(loops): Add OODA loops MCP integration

# Bug fix
fix(analytics): Resolve concurrent write issue in loop-usage.jsonl

# Breaking change (note the !)
feat!: Remove deprecated thoughtbox_v1 tool

# With scope
perf(catalog): Cache hot-loops.json for faster sorting

# Documentation (not in changelog)
docs: Update README with loops documentation

Why This Matters

  1. Automated changelog: Commits automatically populate CHANGELOG.md
  2. Semantic versioning: Commit types determine version bumps
  3. Clear history: Easy to see what changed and why
  4. PR quality: GitHub Action validates format

Read the full file on GitHub · 133 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. yesterday First seen · 133 lines · 1,297 tokens per session scan B cc55b2dcd345

Subscribe to this mod's changes

thoughtbox CLAUDE.md is an instructions file published in the GitHub repository Kastalien-Research/thoughtbox (64 stars, last pushed 1mo ago), licensed MIT. It adds 1,297 tokens to every session, about $0.0065 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.