martin-loop
Keesan12/martin-loop
Governed runtime for AI coding agents with budget caps, verifier gates, runaway-loop prevention, rollback evidence and inspectable run receipts.
- Local integrations
Installation, supported tools and required credentials are maintained by the project.
Project details
- GitHub stars
- 414
- Language
- TypeScript
- License
- Apache-2.0
- Forks
- 68
- Latest release
- v0.8.74d ago
- Activity
- updated 2d ago
- ai-agent-runtime
- ai-coding-agents
- ai-governance
- audit-trail
- budget-enforcement
- governed-runtime
- observability
- policy-as-code
First MCP server?Start with the basics
Coding agent that uses language servers for symbolic code operations.
Manage and execute programs, and read, write, search and edit code and text files.
Local agent OS for spec-first AI coding: a replayable, ledger-recorded interview, seed, execute, evaluate and evolve loop that MCP clients can drive.
Index local code into a graph database to give AI assistants context, with graphical code visualizations for humans.
Desktop GUI automation on Windows, macOS and Linux via accessibility APIs instead of screenshots, plus workflow recording, data extraction and DOM inspection.
Coding agent with basic read, write and command line tools.
Deterministic orchestrator for CLI coding agents such as Claude Code, Codex and Aider, with per-agent git worktrees, an audit trail and cost-aware routing.
Framework for building multi-agent systems that coordinate over MCP, with task management, shared context and RAG.