# Mubit

Decision attribution and calibration for AI agents.

Mubit links each outcome to the agent decision that caused it, then uses those findings to change how agents handle similar tasks. Attribute decisions across your existing models and calibrate open-weight models with versioned, reversible adapters.

## Sources
- [Product profile](https://mubit.ai/agents.md): Capture, attribution, calibration, integration and limits.
- [Documentation](https://docs.mubit.ai/)
- [Quickstart](https://docs.mubit.ai/getting-started)
- [Outcome attribution](https://docs.mubit.ai/patterns/outcome-attribution-loop)
- [Provider support](https://docs.mubit.ai/sdk/llm-providers)
- [Framework adapters](https://docs.mubit.ai/sdk/framework-integrations)
- [Coding-agent setup](https://docs.mubit.ai/recipes/agent-memory-skill)
- [Run scopes](https://docs.mubit.ai/concepts/runs-and-scopes)
- [Talk to the team](https://mubit.ai/talk-to-an-engineer/)

Attribution works across models, including GPT and Claude. Adapter calibration is for open-weight models and is versioned and reversible. Homepage illustrations are conceptual, not measured product performance. Confirm commercial terms, supported methods and deployment requirements with the team.