What is claude-bridge CLAUDE.md?

Repository instructions for Claude Code, the coding assistant, covering a Python project that connects Claude to a container through an MCP server. They document the project structure, installation, commands, tests, and debugging.

Understanding the claude-bridge codebase, installing it in editable mode, running its tests, starting the MCP server, and checking the underlying Claude command.

They give the assistant the project-specific context needed to work consistently and run the right commands. This reduces guesswork when changing or testing the repository.

How JosiahSiegel/claude-bridge configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to JosiahSiegel/claude-bridge. 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/JosiahSiegel/claude-bridge/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/JosiahSiegel/claude-bridge

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/josiahsiegel/claude-bridge/claude-md/github.svg)](https://agentmods.dev/instructions/josiahsiegel/claude-bridge/claude-md)

Your own site

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

Per session 4,577 This file is loaded in full into every session.

When invoked 4,577 The same file — it is already loaded in full.

Security scan C 2 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	$0.04577	$0.04577
Opus 5	$0.02289	$0.02289
Sonnet 5	$0.00915	$0.00915
Haiku 4.5	$0.00458	$0.00458

Measured 9d ago against content hash 0390dcd31309, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade C, and why

claude-bridge CLAUDE.md scanned grade C with 2 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 9d 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.

Cloud metadata endpointhighServer-side request forgery

One request to 169.254.169.254 can return temporary IAM credentials.

30. **`notify_url` is validated at create time, not delivery time.** `dispatch_async` and `create_schedule` reject non-`http(s)` schemes, unresolvable hosts, and any address in private / loopback / link-local / multicast

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

23. **Webhooks are fire-and-log, never fire-and-retry.** `_send_webhook` POSTs JSON via `urllib.request` in an executor with a 10s timeout. Successes log `webhook_sent`; failures log `webhook_failed`. Caller is never aff

CLAUDE.md · 91 lines

How it starts

The opening of the file, as written. The whole thing — 91 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 repo is

An MCP server (claude-bridge) that runs inside a devcontainer and lets host-side Claude Cowork dispatch work into the containerized claude -p over docker exec -i stdio. Replaces an earlier file-queue prototype. Python ≥ 3.11, single runtime dependency on mcp (FastMCP), shells out to the claude CLI rather than using the Agent SDK so the bridge inherits whatever auth the user already configured.

Common commands

# install (editable) and run tests
.venv/bin/pip install -e ".[dev]"
.venv/bin/pytest                      # 12 tests, ~3s
.venv/bin/pytest tests/test_dispatcher.py::test_auth_env_passes_through_to_subprocess -v

# run the MCP server interactively (stdio; for debugging via something like
# the MCP inspector, not for normal operation)
.venv/bin/claude-bridge

# sanity-check that the underlying claude works (the bridge will too if this does)
claude -p "hi" --output-format json

There is no lint config and no build step beyond pip install -e.

Architecture

Two files do all the work:

src/claude_bridge/dispatcher.py — Dispatcher class. Spawns claude -p via asyncio.create_subprocess_exec, returns a DispatchResult dataclass. Holds the channel→session-id map, persists it atomically to $HOME/.claude-bridge/sessions.json. Per-channel asyncio.Lock so calls within a channel serialize, but distinct channels run concurrently. Also tracks background Jobs for the async dispatch surface (dispatch_async / get_dispatch / wait_dispatch / cancel_dispatch); job table is bounded by max_completed_jobs (default 1000).
src/claude_bridge/server.py — FastMCP server. Tools split into three groups: synchronous (dispatch), asynchronous long-running (dispatch_async, get_dispatch, wait_dispatch, cancel_dispatch, list_jobs), and channel admin (list_channels, reset_channel). All configuration is env-var-driven (no MCP-side knobs) so the host can't accidentally redirect the state file or change the working directory.

Read the full file on GitHub · 91 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.

9d ago First seen · 91 lines · 4,577 tokens per session scan C 0390dcd31309

Subscribe to this mod's changes

claude-bridge CLAUDE.md is an instructions file published in the GitHub repository JosiahSiegel/claude-bridge (0 stars, last pushed 4mo ago), licensed MIT. It adds 4,577 tokens to every session, about $0.0229 per session on Opus 5. A static security scan graded it C with 2 findings (cloud metadata endpoint, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

next.js AGENTS.md

AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,153 tokens

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

claude-bridge: Instructions file for Claude Code