AKBridge provides three well-structured tools with complete JSON schemas and clear descriptions. Tool naming follows verb_noun convention (search, describe, call). All three tools have descriptions (194-250 chars, within baseline 34-392 range). Input schemas are properly typed with enums, defaults, and constraints. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions are present but generic (e.g., 'Function name or unique alias' lacks format/constraint details). Error handling is minimal, no recovery guidance or actionable error messages visible in the code. Tool composition is sound (search → describe → call workflow), but the 'akbridge_call' tool's output_mode enum and page/page_size parameters lack detailed guidance on when to use each mode.

• Document output schemas for all three tools. For akbridge_search, specify: {results: [{name, alias, category, description}], total_count}. For akbridge_describe, specify: {name, display_name, parameters: {name, type, description, required}, documentation}. For akbridge_call, document each output_mode (raw returns raw data, compact returns summary, summary returns key metrics).
• Add error recovery guidance. E.g., 'If function not found, call akbridge_search() with a partial name. If rate limited, retry after 60 seconds. If invalid arguments, call akbridge_describe() to review parameter schema.'
• Expand parameter descriptions with constraints and examples. E.g., query: 'Search term (1-100 chars, supports function names like "stock_zh_a_hist", categories like "stock", or keywords like "historical price")'. output_mode: 'raw=full API response, compact=summary with key fields, summary=aggregated metrics only'.
• Document pagination behavior. E.g., 'limit parameter caps results per call (default 8, max 50). To fetch more, increment the implicit offset or use cursor if available. Total result count is returned to guide iteration.'
• Add tool annotations to guide LLM behavior. All three tools are read-only (readOnlyHint=true, destructiveHint=false, idempotentHint=true). akbridge_call may have side effects if the underlying AKShare function modifies state, clarify this in the tool description and annotations.

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

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