Summary

DashClaw provides decision infrastructure for AI agents, enabling interception of actions, policy enforcement, human approvals, and audit-ready decision trails. It safeguards against destructive agent actions and ensures compliance, targeting developers building and deploying AI agents.

How It Works

DashClaw acts as an intermediary between AI agents and external systems, evaluating policies before agent actions execute. It leverages various integration methods—MCP Server, platform skills, code hooks, SDKs, and framework plugins—to intercept decisions, apply declarative guard policies, and record verifiable evidence of every interaction. This approach ensures control and auditability over non-deterministic agent behavior.

Quick Start & Requirements

• Installation: Node.js (npm install dashclaw), Python (pip install dashclaw). For Claude Code hooks: npm run hooks:install.
• Deployment: Free tier deployment available on Vercel + Neon.
• Prerequisites: Node.js, Python, npm. Discord account for mobile approvals. Optional: Upstash Redis for live event streaming.
• Demos & Docs: Try the zero-setup Decision Replay demo (npx dashclaw-demo). Integration guides and walkthroughs are linked within the README.

Highlighted Details

• Broad Integration: Supports LangChain, CrewAI, OpenAI, Anthropic, AutoGen, Claude Code, Gemini CLI, and custom agents.
• Flexible Governance: Offers five integration options: MCP Server, Platform Intelligence Skill, Claude Code hooks, SDK, and OpenClaw plugin.
• Multi-Channel Approvals: Facilitates human-in-the-loop (HITL) approvals via Discord (mobile PWA), CLI, Telegram, and the Mission Control dashboard.
• Comprehensive Features: Includes Mission Control dashboard, Approval Queue, Guard Policies, Drift Detection, Prompt Injection Scanning, Capability Registry, Workflow Engine, Agent Profiles, and detailed Analytics.
• Setup Assistance: The doctor tool diagnoses and auto-fixes common setup issues.

Maintenance & Community

The project is built by Practical Systems. No specific details on community channels (Discord, Slack) or active maintainer information were found in the provided text.

Licensing & Compatibility

Licensed under MIT. No explicit restrictions for commercial use or closed-source linking were noted.

Limitations & Caveats

The "Free while we grow" model indicates a future Pro tier launch based on integration metrics. Hosted trial deployments require additional environment variables and operator-managed infrastructure. Runtime components remain free for solo developers.