Build Strategy to ensure accurate retrieval from MCP server.

[Shubham-Khichi]

ok so one of the challenges I am facing is that whenever I pull entire documentation from a source like https://docs.ag2.ai/docs and DevDocs adds the data into the MCP server. Claude (via Roo Cline) has a hard time digesting ALL the data together due to the context window and you will see an error like this

400 {"type":"error","error":{"type":"invalid_request_error","message":"prompt is too long: 208062 tokens > 200000 maximum"}}

To counter this we have created some inbuilt micro strategy which are designed to do only one thing. Take your query via Cline/Roo Cline chat and search for answers by strategically navigating the entire documentation source of the MCP server. This way even if you pull entire documentation source of AWS or Microsoft you dont have to worry about maxing out your token as DevDocs will only search for the answer to the query hereby lowering retrieval times and token.

I will add this feature shortly to the Main branch.