# OpenWorkProof

> OpenWorkProof is an open-source MCP server for security. Verify an AI agent's signed work records and evidence bundles; includes a SOC review demo. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, self-hosted
- **Repository:** https://github.com/dengyier/openworkproof
- **MCP Registry name:** io.github.dengyier/OpenWorkProof
- **Categories:** [security](https://ai-powered.com/categories/security), [AI models and LLM tooling](https://ai-powered.com/categories/ai-and-llm)
- **Listing updated:** 2026-11-18

## About

OpenWorkProof is aimed at teams that want to check what an agent was permitted to do, what it recorded and whether that evidence holds up. It does not watch every agent on your machine; your integration has to supply the records. You keep the final accept or reject decision.

The server exposes 32 tools over stdio. The README suggests starting with a synthetic run: fetch sample alerts, analyze them as Slips alert cases, then verify the returned bundle. Verification stays UNKNOWN unless you obtain trusted keys and checkpoints separately, which is expected rather than an installation failure. The SOC workflow uses a deterministic investigation stub and simulated actions, so it does not test detection accuracy or remediate hosts.

Separately, an experimental Guard lets Codex edit supported text files inside one preview directory. It must be enabled on its own, the recorded baseline is macOS with Codex CLI 0.147.0, and its approval prompts are currently in Chinese. No OWP cloud account or API key is needed; install with uvx.

## Features

- **Signed work records:** Check task scope, recorded actions and evidence in an agent's records.
- **Evidence bundles:** Export bundles another reviewer can inspect, with offline verification commands.
- **SOC alert review:** Group supplied Slips alerts into cases and export a signed bundle.
- **Codex Guard:** Experimental preview-directory limits for Codex edits.
- **CLI verifiers:** Commands for surface, acceptance and SOC bundles.
- **Optional skill:** A SOC review procedure that agents can install.

## Good for

- Check that exported agent evidence is intact
- Review simulated SOC alert decisions
- Limit a Codex edit task to a preview directory
- Add evidence verification to a GitHub Action
- Hand a reviewer a verifiable bundle

## Install

### Claude Code (Terminal)

```sh
claude mcp add openworkproof -- uvx openworkproof
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "openworkproof": {
      "command": "uvx",
      "args": [
        "openworkproof"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "openworkproof": {
      "command": "uvx",
      "args": [
        "openworkproof"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "openworkproof": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "openworkproof"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "openworkproof": {
      "command": "uvx",
      "args": [
        "openworkproof"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "openworkproof": {
      "command": "uvx",
      "args": [
        "openworkproof"
      ]
    }
  }
}
```

## Tools

- `owp_soc_sample_alerts`: Return a synthetic sample of SOC alerts
- `owp_soc_analyze_slips`: Analyze Slips alerts into review cases and an evidence bundle
- `owp_soc_verify_bundle`: Verify a returned evidence bundle

## FAQ

### Do I need an account or key?

No. It requires no OWP cloud account or API key.

### Why is my verification result UNKNOWN?

It is expected when independent trusted keys and checkpoints are missing.

### Is Guard ready for production?

The README labels it experimental and separately enabled.

### Which Python is needed?

Python 3.10 or later, plus uv.

### Is OpenWorkProof free?

Yes. OpenWorkProof is open source, so you can use and self-host it for free.

### How do I add OpenWorkProof to Claude Code?

Run this command in your terminal: claude mcp add openworkproof -- uvx openworkproof

### What are the best alternatives to OpenWorkProof?

The closest alternatives to OpenWorkProof are REA, headroom, omniget, next-ai-draw-io and mcp-for-blender.

## Where this comes from

- Description: written from the project's README.
- Install steps: from the official MCP Registry entry.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [headroom](https://ai-powered.com/mcp/headroom-3): headroom is an open-source MCP server for ai models and llm tooling.
- [omniget](https://ai-powered.com/mcp/omniget): omniget is an open-source MCP server for ai models and llm tooling.
- [next-ai-draw-io](https://ai-powered.com/mcp/next-ai-draw-io-2): next-ai-draw-io is an open-source MCP server for ai models and llm tooling.
- [mcp-for-blender](https://ai-powered.com/mcp/mcp-for-blender): mcp-for-blender is an open-source MCP server for design and media.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [unity-mcp](https://ai-powered.com/mcp/unity-mcp-2): unity-mcp is an open-source MCP server for ai models and llm tooling.
- [fastapi_mcp](https://ai-powered.com/mcp/fastapi-mcp): fastapi_mcp is an open-source MCP server for ai models and llm tooling.

---
Source: https://ai-powered.com/mcp/openworkproof
