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/pepperonas/pigeon/claude-mdgit clone --depth 1 https://github.com/pepperonas/pigeonWrote 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/pepperonas/pigeon/claude-md)<a href="https://agentmods.dev/instructions/pepperonas/pigeon/claude-md"><img src="https://agentmods.dev/badge/instructions/pepperonas/pigeon/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 5.1 | $0.03045 | $0.03045 |
| Opus 5 | $0.01522 | $0.01522 |
| Sonnet 5 | $0.00609 | $0.00609 |
| Haiku 4.5 | $0.00304 | $0.00304 |
Grade A, and why
pigeon 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 167 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.
What this is
Pigeon forwards browser console errors (and uncaught exceptions, unhandled rejections, failed
network requests) to Claude Code over MCP, and can drive the browser back. A Chrome MV3
extension captures errors and ships them over a WebSocket to a standalone bridge daemon;
each Claude Code session runs a thin MCP proxy that talks to the daemon over a control
channel — so many sessions/projects share one browser feed. The daemon also serves a local
web dashboard and backs a pigeon CLI (status/doctor/stop/dashboard).
page → injected(MAIN) → content(ISOLATED) → service worker → WS(:8765) → bridge daemon
│ control channel (:8766)
MCP proxy (per session) ┤ dashboard http (:8767)
pigeon CLI ┴─ stdio ─→ Claude Code
Commands
npm run install:all # install server/ + extension/ deps
npm run build # build both packages
# server/ (tsc, Node16 ESM)
npm --prefix server run build
npm --prefix server start # MCP proxy (how Claude Code launches it; auto-spawns the daemon)
npm --prefix server run bridge # run the daemon directly (debugging)
npm --prefix server run cli -- status # pigeon CLI: status | doctor | stop | dashboard
npm --prefix server run test:e2e # full proxy/daemon E2E (drives every tool over the control channel)
npm --prefix server run test:multi # two proxies share one daemon + pageUrl scoping + session identity
npm --prefix server run test:ports # daemon shifts ports when 8765/8766 are taken
npm --prefix server run test:dashboard # daemon's dashboard HTTP API: token gating, /api/state, clear
npm --prefix server run test:sourcemap # source-map resolution test
# extension/ (esbuild bundle + tsc typecheck)
npm --prefix extension run build
npm --prefix extension run typecheck
npm --prefix extension run test:unit # pure serialization unit tests (esbuild + node:assert)
npm --prefix extension run test:browser # loads the extension into Chrome for Testing (real E2E)
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.
- 4d ago First seen · 167 lines · 3,045 tokens per session scan A 3b736dfb384b
pigeon CLAUDE.md is an instructions file published in the GitHub repository pepperonas/pigeon (0 stars, last pushed 3mo ago), licensed MIT. It adds 3,045 tokens to every session, about $0.0152 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.
Other instructions, from other repositories
css-noop-checker CLAUDE.md
Instructions for purupurupu/css-noop-checker, covering claude.md, project overview, commands, tech stack and architecture.
css-noop-checker AGENTS.md
Instructions for purupurupu/css-noop-checker, covering ai agent instructions, project structure, adding a new rule, 1. create the rule file and 2. register in engine.ts.
underpixel CLAUDE.md
Claude Code instructions for Leluth/underpixel, covering underpixel, commands, workflow, non-obvious conventions and design docs.
agentclip AGENTS.md
AGENTS.md instructions for 0xkaz/agentclip, covering agents.md — agentclip, project, stack, layout and commands.
agentclip CLAUDE.md
Claude Code instructions for 0xkaz/agentclip, a project described as: Clip browser text, recall it from any AI — Chrome extension + Cloudflare Workers with MCP, semantic search & per-clip encryption.
dev3000 AGENTS.md
Instructions for vercel-labs/dev3000, covering agents.md, runtime, local ui, production workflows, browser tools and development rules.