Three tools with complete input schemas and descriptions, but significant gaps in output documentation and parameter clarity. All tools are read-only GET operations with good naming (verb_noun pattern). Schemas are present and properly typed via schemars JsonSchema derivation. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions are verbose but sometimes unclear (e.g., 'Filter incomplete cards' lacks context on what 'incomplete' means). No error handling guidance. Tool descriptions are adequate (100-150 chars) but lack WHEN/WHY context. Missing pagination guidance despite API returning lists.

• Document output schema for each tool. Define the structure of the JSON/text response (e.g., array of card objects with fields: id, name, cost, type, traits, etc.). This enables LLMs to extract and chain data.
• Add pagination parameters (limit, offset or cursor) to all three tools and document max result limits. Cap results at 20-50 items per call to prevent context window exhaustion.
• Clarify parameter descriptions with domain context. E.g., 'incomplete: boolean, if true, include cards with missing artwork or rules text; if false (default), only complete cards.' Replace format examples with enum constraints where possible.
• Enhance tool descriptions with WHEN/WHY guidance. E.g., 'Use to search for specific cards by name, cost, type, or traits. Call this before get_packs if you need to find which pack contains a card.'
• Add error recovery hints in tool descriptions or via structured error responses. E.g., 'If the API is unavailable, retry after 5 seconds. If a filter value is invalid, check the documentation for allowed values.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry