RSS Feed Scraper has complete JSON schemas with proper types and descriptions for all 5 tools. Tool names follow verb_noun convention (list, get, download, transcribe, update). However, descriptions are minimal (10-50 chars), parameter descriptions lack context about constraints and formats, and error handling is absent. No tool annotations (readOnlyHint/destructiveHint). Output schemas are documented but lack pagination support for list operations. The server lacks guidance on when to use each tool and what to do on failure.

• Expand tool descriptions to 100-200 chars. Example: 'listFeeds: Retrieve all configured RSS feed sources (name and URL). Call this first to discover available feeds before fetching content.' This tells the LLM what the tool does, when to use it, and what it returns.
• Add parameter descriptions with constraints. Example: 'budget: Maximum USD to spend on Whisper transcription (0.01 - 100.00). If exceeded, processing stops and returns partial results.' Include units, ranges, and defaults.
• Add readOnlyHint and destructiveHint annotations to tool definitions in mcp.json. Mark listFeeds and getFeedContent with readOnlyHint=true; mark downloadEpisode, transcribeEpisode, updateDatabase with destructiveHint=true.
• Implement pagination for getFeedContent: add limit (default 20, max 100) and offset parameters. Return total count and next_offset in result. This prevents context window exhaustion on large feeds.
• Add error recovery guidance in tool descriptions. Example: 'If transcription fails due to API rate limits, wait 60 seconds and retry. If the audio file is corrupted, re-download using downloadEpisode.' Include actionable next steps for common failures.