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/linxule/deepthonk/claude-mdgit clone --depth 1 https://github.com/linxule/deepthonkWrote 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/linxule/deepthonk/claude-md)<a href="https://agentmods.dev/instructions/linxule/deepthonk/claude-md"><img src="https://agentmods.dev/badge/instructions/linxule/deepthonk/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 | $0.03509 | $0.03509 |
| Opus 5 | $0.01754 | $0.01754 |
| Sonnet 5 | $0.00702 | $0.00702 |
| Haiku 4.5 | $0.00351 | $0.00351 |
Grade C, and why
deepthonk CLAUDE.md scanned grade C 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 3d 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
- **Anything `test/` imports must be declared in the root `package.json`.** `test/` lives at the workspace root, so a package that only reaches it transitively resolves fine locally and then fails CI with `ERR_MODULE_NOT How it starts
The opening of the file, as written. The whole thing — 123 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Project
DeepThonk implements OpenDeepThink as a provider-neutral TypeScript CLI + MCP server. A single shared engine in packages/core drives a population-based reasoning loop: generate → pairwise judge → Bradley-Terry rank → critique-guided mutation with elite preservation → final dense ranking. See AGENTS.md for the build contract and docs/algorithm.md for the loop details.
Toolchain
Uses pnpm workspaces (not bun) — pinned to [email protected] via the root packageManager field (pnpm 11 requires Node ≥ 22.13 and is needed for OIDC trusted publishing). Node ESM ("type": "module"). TypeScript 7 with project references; tests use Vitest 4 with path aliases that resolve @deepthonk/* to packages/*/src so tests run against source, not built dist/. pnpm-workspace.yaml sets allowBuilds: { esbuild: true } (pnpm 11 errors on unapproved build scripts). CI runs Node 22 and 24; @types/node stays on ^22 to match the floor.
Commands
pnpm install
pnpm run build # tsc per workspace package
pnpm test # vitest run (all packages)
pnpm run lint # typecheck only — no separate linter
pnpm run typecheck # same as lint
# Single test file or test
pnpm exec vitest run test/core/runner.test.ts
pnpm exec vitest run -t "name pattern"
# Per-package build/typecheck
pnpm --filter @deepthonk/core run build
pnpm --filter deepthonk run typecheck
# CLI (from clone, --silent keeps stdout JSON-parseable)
pnpm --silent --filter deepthonk deepthonk plan --profile paper
pnpm --silent --filter deepthonk deepthonk run \
--provider fake --profile quick \
--task examples/tasks/toy-math.txt --out runs/test-quick
pnpm --silent --filter deepthonk deepthonk inspect runs/test-quick
# Dev CLI without building
pnpm --filter deepthonk run deepthonk -- plan --profile quick
The CLI also exposes dt as a short alias once installed. Built bin is packages/cli/dist/index.js.
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.
- 3d ago First seen · 123 lines · 3,509 tokens per session scan C b3d05c71f275
deepthonk CLAUDE.md is an instructions file published in the GitHub repository linxule/deepthonk (0 stars, last pushed 3d ago), licensed MIT. It adds 3,509 tokens to every session, about $0.0175 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
tuui AGENTS.md
Instructions for AI-QL/tuui, covering agents guidelines for this repository, 1. use the development server, not npm run build, 2. keep dependencies in sync, 3. coding conventions and 4. project structure.
airelay CLAUDE.md
Claude Code instructions for ligj1706/airelay, covering claude.md, 项目概述, 产品定位, 设计原则 and 产品演进方向.
hoop CLAUDE.md
Claude Code instructions for hoophq/hoop, covering claude.md, project overview, toolchain & prerequisites, architecture and module breakdown.
gini-agent AGENTS.md
AGENTS.md instructions for Open-Curiosity/gini-agent, covering gini agent instructions, shape, adrs, boundaries and branches.
InvestSkill GEMINI.md
Gemini CLI instructions for yennanliu/InvestSkill, covering investskill — gemini cli setup & usage guide, installation & setup, quick start, navigate to the investskill directory and start gemini cli (loads gemini.md automatically).
rome CLAUDE.md
Claude Code instructions for rome-os/rome, covering rome, playbook and traps.