[P] Easily provide Wandb logs as context to agents for analysis and planning.

It is frustrating to use the Wandb CLI and MCP tools with my agents. For one, the MCP tool basically floods the context window and frequently errors out :/

So I built a cli tool that:

• imports my wandb projects;
• uses algorithms from AlphaEvolve to index and structure my runs;
• is easy to use for agents;
• provides greater context of past experiments;
• does not flood the context window; and
• easily tune exploration-exploitation while planning

Would love any feedback and critique from the community :)

Repo: https://github.com/mylucaai/cadenza

Along with the cli tool, the repo also contains a python SDK which allows integrating this into other custom agents.