| FMRS | 43 / 100 · D | 79 / 100 · B |
|---|
| Reliability | 4 / 20 | 12 / 20 |
|---|
| Security and permissions | 8 / 20 | 18 / 20 |
|---|
| Maintenance | 8 / 20 | 16 / 20 |
|---|
| Documentation | 13 / 20 | 18 / 20 |
|---|
| Setup experience | 10 / 20 | 15 / 20 |
|---|
| Best for | - Developers who regularly use Claude Code, Codex CLI, Gemini CLI, or other supported AI coding CLIs.
- Users seeking to reduce context usage from large repositories, logs, screenshots, and skill files.
- Users comfortable with modifying local AI-client configuration and creating indexes and caches.
| - 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 | - Users seeking a standard MCP server manifest, documented MCP transport, or directly callable MCP tool inventory.
- Users who do not want a global npm package, registered hooks, or client-configuration changes.
- Production users requiring validated macOS support; the README labels macOS as untested.
| - 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 | - Requires installing a global Node.js/npm package.
- Requires writing client configuration and hook files for supported AI tools.
- Requires reading project files, Git information, and relevant session data to build indexes, caches, and compact manifests.
- The optional worker runs as a local background process.
| - 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 | - Hooks intercept or rewrite reads, Bash commands, screenshots, and other tool calls, which can change how original output is presented.
- Indexes, caches, and session manifests store project or tool output locally; the README does not document complete retention or encryption behavior.
- The README claims to scan fetched web content and block prompt injection, but this should not replace human review or the client's own security controls.
- Some bridges are not validated against live instances, including the documented Qwen, OpenClaw, and pi integrations; Copilot hook failures can deny all tool calls in a session.
- Repository metadata says NOASSERTION for the license, while the README badge shows PolyForm Noncommercial; verify the repository license before use.
| - 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, Qwen Code, GitHub Copilot CLI, OpenCode, pi, Grok CLI | Claude Code, Codex, Gemini CLI, Cursor, VS Code, GitHub Copilot, Continue |
|---|
| Tools | 0 | 0 |
|---|