DriveMem exposes 5 tools with complete input schemas and descriptions. Naming follows verb_noun convention (aidrive_search, aidrive_ask, aidrive_store, aidrive_harvest, aidrive_get_context). Descriptions are substantive (100-250 chars) and explain WHEN to use each tool. However, output schemas are not documented, LLMs cannot predict response structure for downstream chaining. Parameters lack type constraints (e.g., contextBudget is 'number' but no min/max bounds). Error handling and recovery guidance are absent. Tool composition is sound (search → ask → store → harvest pipeline), but missing idempotency hints and confirmation patterns for write operations (store, harvest).

• Document output schemas for all tools. Example for aidrive_search: {type: 'object', properties: {results: {type: 'array', items: {type: 'object', properties: {file_id, title, excerpt, relevance_score}}}, total_count, next_cursor}}. This enables LLMs to chain results into downstream tools.
• Add min/max constraints to numeric parameters. E.g., contextBudget: {type: 'number', minimum: 100, maximum: 50000, description: '...token budget (100-50000, default 8000)'}. Prevents absurd values.
• Add idempotency and confirmation hints to aidrive_store and aidrive_harvest. E.g., 'This operation is idempotent, calling twice with the same content produces one entry, not duplicates.' Or offer a dry_run parameter to preview changes.
• Document error scenarios and recovery paths. E.g., 'If aidrive_search returns no results, try aidrive_ask with a rephrased question or broader scope. If aidrive_store fails with 'quota exceeded', contact support or delete old entries.'
• Standardize parameter names and descriptions to English. Keep Chinese translations in separate localization files, not in the schema. This improves LLM parsing and reduces confusion.