What is comet-mcp CLAUDE.md?

A project guide for Comet MCP, a server that connects Claude Code to Perplexity's Comet browser through Chrome DevTools Protocol, a way for software to control a browser. It describes the server's six browser and research tools.

Use it when changing Comet connection, question, polling, stopping, screenshot, or search-mode features, or when building and testing the MCP server.

The guide explains how the connection works and how responses are detected, which helps avoid incorrect changes to this specialised integration. It also documents how to build and restart it.

How hanzili/comet-mcp configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to hanzili/comet-mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/hanzili/comet-mcp/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/hanzili/comet-mcp

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/hanzili/comet-mcp/claude-md/github.svg)](https://agentmods.dev/instructions/hanzili/comet-mcp/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/hanzili/comet-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/hanzili/comet-mcp/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/hanzili/comet-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/hanzili/comet-mcp/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 530 This file is loaded in full into every session.

When invoked 530 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00530	$0.00530
Opus 5.5	$0.00212	$0.00212
Sonnet 5.5	$0.00106	$0.00106
Haiku 4.5	$0.00053	$0.00053

Measured 2mo ago against content hash 5a9a3635d05d, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

comet-mcp 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 2mo 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 · 52 lines

How it starts

The opening of the file, as written. The whole thing — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

What This Is

MCP server connecting Claude Code to Perplexity's Comet browser via Chrome DevTools Protocol (CDP).

Architecture

Claude Code → MCP Server (index.ts) → CometAI (comet-ai.ts) → CDP Client (cdp-client.ts) → Comet Browser

6 Tools

comet_connect - Start/connect to Comet browser
comet_ask - Send prompt, wait for response (15s default, use poll for longer)
comet_poll - Check status of long-running tasks
comet_stop - Stop current task
comet_screenshot - Capture current page
comet_mode - Switch Perplexity modes (search/research/labs/learn)

Key Implementation Details

Response extraction (comet-ai.ts:getAgentStatus):

Takes LAST prose element (not longest) - conversation threads show newest last
Filters out UI text (Library, Discover, etc.) and questions (ends with ?)

Follow-up detection (index.ts):

Captures old prose count/text before sending
Waits for NEW response (different text or more elements)

Prompt normalization:

Strips bullet points, collapses newlines to spaces

Build & Test

npm run build
pgrep -f "node.*comet-mcp" | xargs kill  # Restart MCP

Manual testing only (integration code, external DOM dependency).

Test Cases

Quick queries - Simple questions (math, facts) should return within 15s
Non-blocking - Short timeout returns "in progress", use poll to get result
Follow-up - Second question in same chat detects NEW response correctly
Agentic task - "Take control of browser and go to X" triggers browsing
newChat after agentic - newChat: true resets CDP state after browser control
Mode switching - comet_mode changes search/research/labs/learn

Known Edge Cases

Prompt not submitted: If response shows 0 steps + COMPLETED, prompt may not have been submitted. Retry or use newChat.
Stale poll response: If poll returns unrelated response, the previous prompt failed. Send again.
Research mode: Takes longer than search mode, may need multiple polls.

Read the full file on GitHub · 52 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.

2mo ago First seen · 52 lines · 530 tokens per session scan A 5a9a3635d05d

Subscribe to this mod's changes

comet-mcp CLAUDE.md is an instructions file published in the GitHub repository hanzili/comet-mcp (181 stars, last pushed 9mo ago), licensed MIT. It adds 530 tokens to every session, about $0.0021 per session on Opus 5.5. 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 2026-08-30.

Browser automation agentic-ai agentic-browser automation mcp-server

comet-mcp CLAUDE.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
comet-mcp CLAUDE.mdhanzili/comet-mcp	Instructions file	181	9mo ago	MIT	A	530	Claude Code
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	9,719	GitHub Copilot
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	5,516	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	yesterday	MIT	A	7,461	Codex, OpenCode
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	today	MIT	A	4,680	Codex, OpenCode
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	3,093	GitHub Copilot
vscode model-prompts.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	2,022	GitHub Copilot

comet-mcp: Instructions file for Claude Code