| FMRS | 50 / 100 · D | 79 / 100 · B |
|---|
| Reliability | 4 / 20 | 12 / 20 |
|---|
| Security and permissions | 8 / 20 | 18 / 20 |
|---|
| Maintenance | 15 / 20 | 16 / 20 |
|---|
| Documentation | 14 / 20 | 18 / 20 |
|---|
| Setup experience | 9 / 20 | 15 / 20 |
|---|
| Best for | - Developers already using Claude Code or similar coding agents who want a graphical desktop workspace
- Users who need to manage multiple model providers (Claude/ChatGPT/Grok/local models) and multiple MCP server connections in one place
- Teams or individuals who want to monitor and approve agent actions remotely through chat apps
| - Engineering teams scaling AI coding agents while controlling spend
- Platform teams that need auditable, rollback-aware agent execution records
- Organizations using multiple agents (Claude, Codex, Gemini) under one governed flow
|
|---|
| Not for | - Anyone looking for an MCP server that Claude Desktop, Claude Code, or another MCP client can connect to and call specific tools on
- Headless server deployments where a desktop Electron GUI isn't wanted or usable
- Users unwilling to install a local Electron desktop app or wary of installing skills from a third-party marketplace
| - Developers who just want a simple code-completion helper without run governance
- Teams unwilling to accept local JSONL run records with HMAC signing
- Environments without a supported adapter (Claude/Codex/Gemini CLIs, direct provider, or verifier-only)
|
|---|
| Required permissions | - Local file system access for Git Worktree operations, diff review, and project files
- Network access to reach model provider APIs and IM channels (WeChat, Feishu, DingTalk, Telegram, WhatsApp)
- Computer Use permission (once authorized) allowing the agent to take screenshots, click, type, and control desktop apps
- A 'skip permissions' mode is available among the five permission levels, letting the agent run commands and tool calls without per-action approval
| - Read/write access to the configured repo root (writable scope restrictable via --allow-path/--deny-path)
- Ability to invoke the chosen agent CLI (Claude/Codex/Gemini, etc.) and its underlying model account
- Local disk writes for run records, receipts, and share outputs (share/run-receipt., etc.)
- Execution of user-configured verifier commands (e.g. npm test)
|
|---|
| Risks and side effects | - Computer Use can directly control the desktop once authorized; an overly permissive permission mode raises the chance of unintended actions
- The app stores API keys for multiple model providers locally, concentrating sensitive credentials on one device
- The skill marketplace allows installing third-party skills from ClawHub/SkillHub; source and safety status are shown, but supply-chain risk remains
- Unsigned draft or temporary builds require manual one-time approval to run, which can be risky if the build source isn't trusted
- As an MCP client, it can be pointed at arbitrary third-party MCP servers, so users must independently vet the trustworthiness of any server they add
| - Verifier commands are executed; poorly chosen verifier commands carry security risk despite preflight policy checks
- Cost and token figures may be estimates (labeled with provenance) and should not be treated as settled accounting
- Receipt integrity must be verified before use as trustworthy evidence; missing rollback evidence is flagged as EVIDENCE_BOUNDARY
- Policy defaults come from martin.config.yaml and can be overridden by CLI flags; misconfiguration can loosen safety boundaries
|
|---|
| Supported clients | | Claude Code, Codex, Gemini CLI, Cursor, VS Code, GitHub Copilot, Continue |
|---|
| Tools | 0 | 0 |
|---|