claw-ea CLAUDE.md

claw-ea CLAUDE.md is an instructions file for coding agents from fshaan/claw-ea. It costs 3,591 tokens per session, scanned A, original, MIT.

A set of project instructions for Claude Code, an AI coding assistant, in the claw-ea repository. It explains the project's structure, the boundary between the AI agent and its tools, and its 11 available MCP tools.

In plain words
What is it for?
Use it when working on claw-ea, a server that archives messages and files from communication services into Obsidian, Apple Calendar, and Apple Reminders.
Why use it?
It gives the coding assistant the project context needed to use tools correctly and keep reasoning separate from side effects such as writing files or calling system APIs.

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/fshaan/claw-ea/claude-md
Clone the repo
git clone --depth 1 https://github.com/fshaan/claw-ea

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 claw-ea CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/fshaan/claw-ea/claude-md.svg)](https://agentmods.dev/instructions/fshaan/claw-ea/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/fshaan/claw-ea/claude-md"><img src="https://agentmods.dev/badge/instructions/fshaan/claw-ea/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 3,591 This file is loaded in full into every session.
When invoked 3,591 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 $0.03591 $0.03591
Opus 5 $0.01795 $0.01795
Sonnet 5 $0.00718 $0.00718
Haiku 4.5 $0.00359 $0.00359

Measured 3d ago against content hash b699933b817c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

claw-ea 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 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.

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 · 246 lines

How it starts

The opening of the file, as written. The whole thing — 246 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 Overview

claw-ea is a Python MCP (Model Context Protocol) server that helps medical professionals automatically archive messages from social media channels (Feishu, WeCom, Telegram, etc.) into Obsidian notes, Apple Calendar, and Apple Reminders. It runs as an external MCP server connected to OpenClaw via MCPorter, but is designed to work with any MCP client (Claude Desktop, Cursor, etc.).

Architecture

Agent vs Tool Boundary (Critical Design Principle)

The MCP server only provides side-effect tools — tools that write files, call system APIs, or read system state. All "understanding" tasks (message classification, image comprehension, approval dialog formatting) are handled by the calling agent's LLM, NOT by our tools.

MCP Tools (11 total)

Core workflow (called by agent after it classifies the message):

  • ocr_image — Local OCR only (macOS Vision framework). Agent's multimodal LLM handles image understanding directly; this tool is the fallback when the LLM lacks vision.
  • save_attachment — Saves file to date-organized attachment directory. Two modes: file_path (local path, preferred — uses shutil.copy2) or file_content (base64). Dedup: skips if identical file exists.
  • convert_to_markdown — Converts files (PDF, docx, pptx, xlsx, CSV, HTML, images, plaintext) to Markdown via configurable converter chains (docling → markitdown → mineru → lmstudio → vision_ocr → passthrough). Returns temp file path for create_obsidian_note's raw_body_path. Fallback: if primary converter fails or output is garbled, automatically tries the next in chain. Plaintext files (.txt, .md, .rst, .log) use a passthrough converter. File-type routing: images/PDF are NOT routed here by default — the agent reads them directly with its multimodal vision, summarizes, and keeps the original as an embedded attachment (![[file]]); this tool is only a fallback when the agent lacks vision. Office docs docx/pptx/xlsx go through MinerU (offline, local, default; docling fallback); csv/html route to markitdown/docling instead. Other types: agent picks the extraction, but the original is always archived via save_attachment.
  • create_obsidian_note — Creates Markdown note with YAML frontmatter in Obsidian vault. Supports raw_body_path to embed converted Markdown content. Dedup via content hash (first 8 chars in filename); returns existing path if duplicate.
  • create_calendar_event — Creates event in Apple Calendar via pyobjc EventKit. Returns event ID.
  • delete_calendar_event — Deletes an event from Apple Calendar by event ID. Agent should confirm with user first.
  • create_reminder — Creates reminder in Apple Reminders via pyobjc EventKit. Returns reminder ID.
  • delete_reminder — Deletes a reminder from Apple Reminders by reminder ID. Agent should confirm with user first.

Read the full file on GitHub · 246 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. 3d ago First seen · 246 lines · 3,591 tokens per session scan A b699933b817c

Subscribe to this mod's changes

claw-ea CLAUDE.md is an instructions file published in the GitHub repository fshaan/claw-ea (1 stars, last pushed 2mo ago), licensed MIT. It adds 3,591 tokens to every session, about $0.0180 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.