A Flask and Streamlit-based NLP framework for analyzing Lex Fridman podcast episodes. Provides topic modeling (NMF, LSA, LDA), emotion analysis, guest tracking, and a LangChain-powered chatbot with Wikipedia search and topic prediction capabilities.
This server has severe definition quality issues. Tool definitions are scattered across multiple files with inconsistent naming (wiki_tool vs WikiSearch, predict_topic vs PredictTopic). Most critically, NO tool has a visible input schema in the provided code, only parameter names and types are inferred from function signatures. Descriptions exist but are minimal (10-30 chars), well below the 50-200 char baseline for LLM-optimized tools. No output schemas are documented. Error handling is generic ('Error: {str(e)}') with no recovery guidance. The codebase shows Flask routes and Streamlit UI but no MCP server implementation, this appears to be a chatbot application, not an MCP server.
Return known information about a Lex Fridman podcast guest.
Predict the topic of a podcast quote or segment.
Search Wikipedia and return a brief summary of a topic.
Returns a list of all guest names from the dataset.
Predicts the topic of a podcast segment.
Search Wikipedia and return the summary of a topic.
No input schemas visible in source code. Tools lack JSON Schema definitions with type constraints, enums, and parameter descriptions.
Duplicate/conflicting tool definitions. Both 'wiki_tool' and 'WikiSearch' search Wikipedia; both 'predict_topic' and 'PredictTopic' predict topics; both 'list_all_guests' and 'ListAllGuests' list guests. LLMs cannot disambiguate between similar tools.
Descriptions are too short (10-30 chars, well below 50-200 char baseline). 'Predicts the topic of a podcast segment' lacks context on WHEN to use it vs other tools, what format input should be, or what the output structure is.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 25 | 2026-07-28+ | v2 |
No output schemas documented. LLMs cannot plan downstream tool calls or extract structured data. E.g., does 'list_all_guests' return an array of strings, objects with guest metadata, or a single concatenated string?
Generic error handling. wiki_tool catches all exceptions and returns 'Error: {str(e)}' with no recovery guidance. LLMs cannot determine if the error is retryable, user-fixable, or fatal.