ratel
The self-improvement layer for AI agents. Ratel reads your agent's traces, finds what fails, recommends fixes and runs evals on real sessions before you ship. Self-hosted: one app, one Postgres. Includes the TypeScript and Python SDK
Score breakdown — how scoring works
-
Maintenance
30 / 30- Pushed within 14 days0 days ago+30
-
Adoption
16 / 25- GitHub stars (no package published — stars weighted fully)566+16
-
Documentation
25 / 25- Comprehensive README (6,000+ chars)10901 chars+10
- Install / setup instructionsyes+6
- Tools / capabilities documentedyes+5
- Code or client-config exampleyes+4
-
Trust signals
9 / 20- OSS license declaredMIT+7
- First-party vendor implementationno+0
- DNS-verified registry namespaceno+0
- Listed in official MCP Registryno+0
- Owned by an organizationyes+2
Install
From source
git clone https://github.com/ratel-ai/ratel No package published to a registry — see the README for setup instructions.
Always review a server's code and required credentials before connecting it to your MCP client. MCP Vetted scores public metadata — it does not audit code for malicious behavior (yet). Methodology →