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/nickromanek/clawbridge/claude-mdgit clone --depth 1 https://github.com/NickRomanek/clawbridgeWrote 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/nickromanek/clawbridge/claude-md)<a href="https://agentmods.dev/instructions/nickromanek/clawbridge/claude-md"><img src="https://agentmods.dev/badge/instructions/nickromanek/clawbridge/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 | $0.08851 | $0.08851 |
| Opus 5 | $0.04425 | $0.04425 |
| Sonnet 5 | $0.01770 | $0.01770 |
| Haiku 4.5 | $0.00885 | $0.00885 |
Grade D, and why
clawbridge CLAUDE.md scanned grade D 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 6d 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.
Harvests environment variableshighData exfiltration
Enumerating or grepping the environment for keys collects credentials unrelated to what the mod says it does.
- Auth: reads API keys from `~/.openclaw/openclaw.json` `env` block, NOT from `.env`. ClawBridge passes `ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, `OPENROUTER_API_KEY` in a minimal env dict when auto-starting (no `os.environ Reads MCP configurationmediumAgent snooping
mcp.json carries server URLs and auth tokens; reading it lets a mod discover and abuse other integrations.
`clawbridge_mcp.py` exposes 15 tools and 1 resource (`clawbridge://status`). Registered in `.mcp.json`. Runs via stdio (default) or HTTP (`--http` flag). Auth token passthrough on all tools. How it starts
The opening of the file, as written. The whole thing — 368 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 is ClawBridge?
ClawBridge is a desktop and browser automation agent with a web dashboard. It orchestrates three AI-powered engines (browser-use, computer-use, OpenClaw) behind a FastAPI server with a WebSocket-driven chat UI. BYOK (bring your own key) — all LLM API keys stay local.
Development Model: Monolith-First
clawbridge.py (~12,900 lines) is the single-file monolith and the only active development target. All changes go here first. It contains the full application: FastAPI routes, engine implementations, dashboard HTML/JS, safety system, personality manager, task orchestrator, SQLite persistence, and inline CSS/JS.
clawbridge/ is a modular package that mirrors the monolith's logic split across files. It exists for reference and testing but lags behind the monolith. The test suite (tests/) imports from the package, not the monolith.
clawbridge_mcp.py is a standalone MCP server that proxies to the monolith's REST API. It has no shared code with the monolith — it's a thin HTTP client wrapper.
Commands
# Run the server (primary way)
python clawbridge.py
# Dashboard at http://localhost:8765
# Run with Docker
docker-compose up
# Install dev dependencies
pip install -r requirements-dev.txt
# Run all tests (unit + integration, against the package)
pytest
# Run a single test file
pytest tests/unit/test_safety.py
# Run a single test by name
pytest tests/unit/test_safety.py -k "test_scan_detects_aws_key"
# Run with coverage
pytest --cov=clawbridge
# Smoke tests (requires running server on :8765)
python tests/smoke_test.py
# Build portable Windows distribution
python build.py
# Build Windows installer (needs Inno Setup ISCC.exe on PATH)
python build.py --inno
# Build macOS app
python build_macos.py --arch arm64
Architecture
Request flow:
Dashboard (inline HTML/JS) ──WebSocket──▶ FastAPI ──▶ TaskManager ──▶ Engine
│
┌───────┼───────┐
▼ ▼ ▼
browser-use computer openclaw
(Playwright) (pyauto (Node.js
+ LLM) gui+UIA) gateway)
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.
- 6d ago First seen · 368 lines · 8,851 tokens per session scan D 933c1536d416
clawbridge CLAUDE.md is an instructions file published in the GitHub repository NickRomanek/clawbridge (24 stars, last pushed 5mo ago), licensed Apache-2.0. It adds 8,851 tokens to every session, about $0.0443 per session on Opus 5. A static security scan graded it D with 2 findings (harvests environment variables, reads mcp configuration). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
ai-agent-automation AGENTS.md
AGENTS.md instructions for vmDeshpande/ai-agent-automation, covering agents.md, repo structure, dev commands, git hooks and code style.
ui-debugger-mcp CLAUDE.md
Claude Code instructions for developerz-ai/ui-debugger-mcp, covering ui debugger mcp, response rules, what this is, reference repos (read for patterns, don't copy blind) and architecture.
Screenhand CLAUDE.md
Claude Code instructions for manushi4/Screenhand, covering claude.md, project, commands, architecture and key entry points.
Screenhand AGENTS.md
AGENTS.md instructions for manushi4/Screenhand, covering agents.md, project, commands, architecture and key entry points.
n8n-skills CLAUDE.md
Claude Code instructions for czlonkowski/n8n-skills, covering claude.md, project overview, repository structure, the 14 skills and 1. n8n expression syntax.
openscreen AGENTS.md
AGENTS.md instructions for getopenscreen/openscreen, covering agents.md, setup commands, development principles, project layout and code style.