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/base-kit/docs-mcp/claude-mdgit clone --depth 1 https://github.com/base-kit/docs-mcpWrote 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/base-kit/docs-mcp/claude-md)<a href="https://agentmods.dev/instructions/base-kit/docs-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/base-kit/docs-mcp/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 | $0.02850 | $0.02850 |
| Opus 5 | $0.01425 | $0.01425 |
| Sonnet 5 | $0.00570 | $0.00570 |
| Haiku 4.5 | $0.00285 | $0.00285 |
Grade A, and why
docs-mcp CLAUDE.md scanned grade A 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.
Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
- 内置模块一律用 `node:` 前缀（`node:fs`、`node:child_process`、`node:path`），避免与同名 npm 包冲突 How it starts
The opening of the file, as written. The whole thing — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.
docs-mcp 工具开发规范
本仓库是 docs-mcp 工具源码：把开源项目官方文档克隆到本地 → 用共享内核（Orama）索引 → 暴露为 MCP stdio server。 本规范面向 开发 docs-mcp 本身（CLI + 内核 + preset 维护）。用 docs-mcp 查文档写业务代码的规范见
templates/consumer/（消费方模板，可由docs-mcp config --with-claude-md输出到目标项目）。
项目定位与架构
docs-mcp/ ← 单 npm 工程（工具源码）
├── package.json ← CLI + 内核 deps 合一
├── bin/docs-mcp.mjs ← CLI 入口（tsx / node strip-types）
├── tsconfig.json ← 编译 src/ → dist/（sourceMap: false）
├── src/
│ ├── cli/ ← CLI 层（commander，9 命令）
│ │ ├── index.ts ← 命令路由
│ │ ├── commands/ ← 9 个子命令
│ │ ├── log.ts / git.ts / mcp-config.ts
│ ├── core/ ← 共享内核
│ │ ├── indexer.ts ← 文件收集 + Markdown 切块 + 清理
│ │ ├── embed.ts ← all-MiniLM-L6-v2（hf-mirror 镜像）
│ │ ├── server.ts ← MCP stdio server
│ │ ├── tools.ts ← 4 工具实现
│ │ ├── manifest.ts / config.ts / types.ts / build-index.ts
│ └── preset/ ← preset schema (zod) + loader
├── presets/ ← 声明性清单（每服务一 json，提交进 git）
├── services/ ← 指令性索引配置（install/add 自动生成）
├── templates/consumer/ ← 消费方规范模板（CLAUDE.md + mcp-refs）
├── data/ ← 索引产物（gitignored，可重建）
├── packages/ ← 拉取的源码（gitignored）
├── dist/ ← 编译产物（gitignored，无 sourceMap）
└── .mcp.json.template ← .mcp.json 模板（`docs-mcp serve __SERVICE__` 可移植块）
双根：APP_ROOT（代码/只读资源，跟随包安装位置）与 DATA_ROOT（用户数据，~/.docs-mcp/，可被 DOCS_MCP_DATA 覆盖）。data/ packages/ services/ 在 DATA_ROOT 下；仓库内的同名目录仅开发期本地样本（gitignored）。升级包只动 APP_ROOT，不碰 DATA_ROOT。
数据流：presets/<svc>.json → install 克隆到 DATA_ROOT/packages/<pkg>/ + 写 DATA_ROOT/services/<svc>.json → build 跑 dist/core/build-index.js <svc> → 索引落到 DATA_ROOT/data/<svc>/。
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 · 154 lines · 2,850 tokens per session scan A 0a449bc6fe91
docs-mcp CLAUDE.md is an instructions file published in the GitHub repository base-kit/docs-mcp (2 stars, last pushed 2mo ago), licensed MIT. It adds 2,850 tokens to every session, about $0.0143 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
source-reading-methodology AGENTS.md
Instructions for itshen/source-reading-methodology, covering 给 ai 的入口, 要用这套方法论干活 and 要改这个仓库本身.
ScriptXeno.github.io CLAUDE.md
Instructions for ScriptXeno/ScriptXeno.github.io, covering claude.md, project overview, commands, architecture and content model: posts, tags, categories, images.
jdocmunch-mcp CLAUDE.md
Claude Code instructions for jgravelle/jdocmunch-mcp, covering jdocmunch-mcp, standing operational notes (jdoc-specific), standing lessons (suite-wide), release: the two steps that are only written down here and 1. read ci for the pushed sha before any irreversible step.
docs AGENTS.md
Instructions for argos-ci/docs, covering agents.md, always use the documentation skills and keep the table of contents in sync.
libragen AGENTS.md
Instructions for libragen/libragen, covering agent guidelines for libragen, repository structure, package dependencies, key architectural concepts and library manager (packages/core/src/manager.ts).
coven-docs copilot-instructions.md
Copilot instructions for OpenCoven/coven-docs, covering coven-docs, setup and structure.