<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> claw-lens provides an open-source, local observability dashboard tailored for OpenClaw AI agents. It addresses the unique needs of founders, analysts, and operators deploying these agents by offering cost analytics, live monitoring, and debugging tools, moving beyond traditional observability focused on latency and requests. The dashboard prioritizes cost as the dominant signal and sessions as the atomic unit, providing insights into agent behavior and security vulnerabilities that traditional tools miss.

How It Works

This tool operates entirely locally, reading directly from OpenClaw's JSONL logs and workspace files, parsing them into a local SQLite database. An Express API serves a React frontend, ensuring no data leaves the user's machine. Its architecture emphasizes zero configuration, cost-first analysis (USD cost per session, model, agent), local-only operation (read-only access, no outbound calls), and transparent, rule-based security auditing for events like file access, shell commands, and prompt injection.

Quick Start & Requirements

Installation is straightforward via npm: npm install -g claw-lens-cli or by running npx claw-lens-cli for immediate use. The tool requires Node.js 18+. The OPENCLAW_HOME environment variable can specify the data directory. For development, clone the repository, install dependencies (npm install), and run npm run dev. Official documentation is available at claw-lens.com/monitoring/overview and CLI reference at claw-lens.com/reference/cli.

Highlighted Details

• Comprehensive KPI strip (cost, tokens, sessions, errors, cache efficiency) with 7-day trends and week-over-week deltas.
• Detailed 4-dimension cost breakdown (input, output, cache read/write) by agent, model, and time, alongside cache hit rates.
• Real-time agent activity feed via WebSocket proxy to the OpenClaw Gateway.
• Security audit featuring rule-based risk scoring for file access, shell commands, external HTTP calls, sensitive data exposure, prompt injection detection, and per-agent behavioral anomaly detection.
• Session timeline, profiler, context breakdown visualization, and cache trace replay functionality.
• All data remains local; no external data transmission or modification of agent files.
• Supports both English and Chinese interfaces.

Maintenance & Community

No specific details regarding notable contributors, sponsorships, or community channels (e.g., Discord, Slack) were found in the provided README.

Licensing & Compatibility

The project is released under the permissive MIT License, allowing for broad use, modification, and distribution, including within commercial or closed-source applications. No specific compatibility blockers or integration notes beyond its reliance on OpenClaw were mentioned.

Limitations & Caveats

Functionality, particularly cache trace replay, is dependent on OpenClaw being configured to generate the necessary logs. The tool's scope is limited to OpenClaw agents, and its maturity level beyond core features is not explicitly detailed.