What is BrowserMesh CLAUDE.md?

Project instructions for BrowserMesh, a local runtime that manages separate browser sessions for external MCP clients such as Claude Code or Codex.

Developing or operating BrowserMesh while preserving explicit session and page addressing, separated browser-engine adapters, and the project’s architectural boundaries.

They clarify which component is responsible for reasoning and orchestration, and which component only performs browser work.

BrowserMesh CLAUDE.md — Instructions file by scrollDynasty

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/scrolldynasty/browsermesh/claude-md

Clone the repo

git clone --depth 1 https://github.com/scrollDynasty/BrowserMesh

Per session 1,713 This file is loaded in full into every session.

When invoked 1,713 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.

Model	Per session	Once invoked
Fable 5	$0.01713	$0.01713
Opus 5	$0.00856	$0.00856
Sonnet 5	$0.00343	$0.00343
Haiku 4.5	$0.00171	$0.00171

Measured 2d ago against content hash 4618de87907e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

BrowserMesh 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 2d 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 · 136 lines

How it starts

The opening of the file, as written. The whole thing — 136 lines — stays where its author put it; the contents beside it link to each section on GitHub.

BrowserMesh

Local isolated multi-session browser runtime, driven by external MCP clients.

AGENTS.md is the full charter. This file is the working subset; when the two disagree, AGENTS.md wins.

Responsibility boundary

User → external AI client → MCP → BrowserMesh → isolated browser sessions

The external client (Claude Code, Codex, Cursor, Qwen, …) does the reasoning, planning, and orchestration. BrowserMesh only executes browser work.

Never add internal Agent entities or registries, browser_agent_* / browser_message_* tools, agent mailboxes or messaging, LLM-owned session principals, internal LLM calls, or reasoning loops inside BrowserMesh. Session name and metadata are neutral workflow labels — not principals, permissions, identities, or leases.

Architecture invariants

BrowserMesh is an independent runtime, not a wrapper around Playwright MCP. MCP is a transport adapter; Playwright is a browser-engine adapter. Domain, application, and runtime code import neither.

Explicit addressing. Every browser operation targets an explicit sessionId; every page-specific operation an explicit pageId. There is never a global currentSession, activeSession, currentPage, activePage, or currentTab.
Context isolation. Each ready session owns its own non-persistent Chromium BrowserContext.
Page isolation. A pageId from another session is rejected.
Per-session serialization. Everything touching one session's live browser state goes through that session's serial queue — including read-style operations (snapshot, URL, title, visible text). They must not bypass an in-progress operation.
Cross-session parallelism. Different sessions run concurrently. Never introduce one global browser-operation mutex.
Queue recovery. A rejected, failed, or timed-out operation must not poison a session queue; subsequent accepted operations still execute.
Lifecycle safety. Close and shutdown must not race initialization into leaked contexts or pages. An unexpected Chromium disconnect must not silently rebuild live sessions.

Read the full file on GitHub · 136 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.

2d ago First seen · 136 lines · 1,713 tokens per session scan A 4618de87907e

Subscribe to this mod's changes

BrowserMesh CLAUDE.md is an instructions file published in the GitHub repository scrollDynasty/BrowserMesh (4 stars, last pushed 2d ago), licensed Apache-2.0. It adds 1,713 tokens to every session, about $0.0086 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.

Browser automation Data and AI LLM Planning ai-agent ai-agents browser session browser-agent browser-automation

BrowserMesh CLAUDE.md

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

BrowserMesh

Responsibility boundary

Architecture invariants

What this file has done since we first saw it

Other instructions, from other repositories

browser-harness CLAUDE.md

eric-task-master AGENTS.md

AgentLimb AGENTS.md

AgentLimb CLAUDE.md

obscura AGENTS.md

stealth-browser-mcp AGENTS.md