Two tools with complete input schemas and descriptions. wolfram_query has 14 well-documented optional parameters with clear purposes (units, currency, timezone, etc.), meeting the baseline of 4 params per tool. Descriptions are adequate (194 - 250 chars, within the 10 - 1024 baseline) and explain WHAT the tool does. However, output schemas are completely undocumented, LLMs cannot plan downstream calls or extract structured data. No error handling guidance (e.g., 'If query times out, try reducing maxchars'). Parameter descriptions lack format constraints (e.g., 'timezone' should specify IANA format). No mention of rate limits or API key requirements in tool descriptions. wolfram_query_with_assumptions is a composition issue: it duplicates wolfram_query logic rather than being a natural follow-up pattern.

• Document output schema for both tools. Example: 'Returns {success: boolean, result: string, assumptions?: string[], pods?: {title: string, subpods: {plaintext: string}[]}}'. This enables LLMs to extract structured data.
• Add error handling guidance to descriptions: 'If query times out, reduce maxchars or increase totaltimeout. If no results, try simplifying the query to keywords.'
• Specify format constraints in parameter descriptions: timezone='IANA timezone (e.g., America/New_York)', countrycode='ISO 3166-1 alpha-2 (e.g., US, FR)', languagecode='ISO 639-1 (e.g., en, fr)'.
• Merge wolfram_query_with_assumptions into wolfram_query by making 'assumption' an optional parameter. Remove the separate tool to reduce LLM confusion.
• Document API key requirement: 'Requires WOLFRAM_ALPHA_APP_ID environment variable. If missing, returns error: "API key not configured."'