Project Information
Category
Description
Local-first, MIT-licensed cost and efficiency profiler for AI agents built on the openai and anthropic Python SDKs (plus litellm). Finds and quantifies wasted LLM spend (duplicate calls, missed prompt caching, context bloat, model overkill, retry loops) and prints a mechanical fix for two of the five patterns. Runs against one local SQLite file, no account or server required.
Project Information
Category
Description
Local-first, MIT-licensed cost and efficiency profiler for AI agents built on the openai and anthropic Python SDKs (plus litellm). Finds and quantifies wasted LLM spend (duplicate calls, missed prompt caching, context bloat, model overkill, retry loops) and prints a mechanical fix for two of the five patterns. Runs against one local SQLite file, no account or server required.