Two tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. web_search has a reasonable description (89 chars) and complete input schema with types and defaults. web_fetch description is adequate (67 chars) but lacks output schema documentation. Neither tool documents what fields are returned, making it difficult for LLMs to plan downstream operations. Parameter descriptions are minimal (e.g., 'prompt' in web_fetch lacks guidance on format or expected content). No error handling guidance, no idempotency markers, and no indication of which operations are safe to retry. Tool names follow verb_noun pattern (positive), but descriptions lack WHEN/WHY context that LLMs need for selection.

• Document output schema for web_search: specify that results are a list of objects with 'title', 'url', 'snippet' fields (types: string, string, string). Include example structure.
• Document output schema for web_fetch: specify return type (markdown string), max length, and whether it includes metadata (e.g., page title, fetch timestamp).
• Expand web_fetch 'prompt' parameter description: 'What information to extract from the page (e.g., "main article text", "contact information", "pricing table"). Max 200 chars.'
• Add error recovery guidance to both tool descriptions: 'If search fails, the LLM should retry with a simpler query or try a different search backend. If fetch fails, verify the URL is valid and accessible.'
• Add tool annotations (readOnlyHint: true) to both tools in the Tool registration to signal they are safe to call repeatedly without side effects.