| FMRS | 67 / 100 · C | 79 / 100 · B |
|---|
| Reliability | 9 / 20 | 12 / 20 |
|---|
| Security and permissions | 14 / 20 | 18 / 20 |
|---|
| Maintenance | 13 / 20 | 16 / 20 |
|---|
| Documentation | 18 / 20 | 18 / 20 |
|---|
| Setup experience | 13 / 20 | 15 / 20 |
|---|
| Best for | - Large, long-lived, multi-language codebases (up to tens of millions of lines) where an AI assistant needs a durable code index
- Privacy-conscious teams that want embeddings and index data to stay entirely on their own machine or infrastructure
- Users of Claude Code, Cursor, VS Code Copilot, or Gemini CLI who want their AI to prefer semantic search and dependency graphs over raw file reads
- Teams running multiple AI agents or collaborators against a shared, coordinated codebase index
| - 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 | - Environments where Docker cannot be installed or run (e.g. locked-down machines or restricted CI)
- Small projects (a few hundred lines) where the overhead of standing up Docker/Qdrant/Ollama outweighs the benefit
- Teams needing a fully managed, production-ready shared cloud index today — SocratiCode Cloud is still in private beta, not generally available
- Environments with strict approval requirements around new local containers, background file watchers, or third-party npm packages
| - 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 | - Read access to the local project's file system for AST-based chunking and indexing
- Ability to start and manage local Docker containers for Qdrant (vector DB) and Ollama (embeddings)
- Local file system write access to persist indexes, graph data, and optional interactive HTML graph output
- File system watch access to detect changes and keep the index incrementally updated
- Outbound network access and API keys (OPENAI_API_KEY, GOOGLE_API_KEY, QDRANT_API_KEY) when cloud embeddings or an external Qdrant instance are configured
| - 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 | - By default it automatically pulls and runs Docker images (Qdrant, Ollama) — requires trusting those images and their runtime behavior
- Switching to OpenAI or Google cloud embeddings sends code snippets to a third-party API, breaking the 'fully local/private' guarantee — evaluate against code confidentiality requirements
- API keys (OPENAI_API_KEY, GOOGLE_API_KEY, QDRANT_API_KEY) are configured as environment variables — protect config files and shell profiles from leaking them
- Misconfigured external/remote Qdrant instances could expose indexed content, including code snippets or sensitive strings
- The project is licensed AGPL-3.0, which has copyleft implications for integration with proprietary or closed-source software
- Any connected MCP client gets full read/write/delete access to indexing, graph, and context-artifact operations — no fine-grained per-tool permission model is described
- The call graph is static-analysis based and cannot see dynamic dispatch, reflection, or framework magic (e.g. dependency injection, decorator-based routing) — 'zero callers' results should be manually double-checked, not trusted blindly
| - 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, VS Code, Cursor, Gemini CLI, Claude Desktop, Windsurf, Cline, Roo Code, Zed, OpenAI Codex CLI, OpenCode | Claude Code, Codex, Gemini CLI, Cursor, VS Code, GitHub Copilot, Continue |
|---|
| Tools | 21 | 0 |
|---|