Server provides 6 tools with basic schemas and descriptions in French. All tools have input schemas with type definitions and required fields clearly marked. Descriptions are present but brief (10-50 chars), below the 50-200 char baseline for LLM optimization. Parameter descriptions exist but lack detail on constraints, formats, and valid ranges. No output schemas documented. Error handling and recovery guidance absent. Tool naming follows verb_noun pattern (list_, get_, create_) which is correct. No tool annotations (readOnlyHint/destructiveHint) despite clear risk classifications in metadata. Missing idempotency guidance for create_entry.

• Expand all descriptions to 50-200 chars. Example: 'list_projects' → 'List all projects accessible to the authenticated user. Returns project UUIDs, names, and metadata. Call this first to discover available projects before querying collections.'
• Document output schemas for all tools. Specify which fields are returned (e.g., list_projects returns [{project_uuid, name, ...}]) so agents can chain tools correctly.
• Add tool annotations: mark list_* tools with readOnlyHint=true, create_entry with destructiveHint=true. This enables agents to reason about safety and idempotency.
• Add min/max constraints to numeric parameters: limit (1-1000), offset (0+). Add format/pattern to locale (e.g., 'ISO 639-1 code like en, fr'). Add schema for create_entry.data object structure.
• Add error handling guidance: 'If create_entry fails with validation error, call get_collection_schema first to verify field names and types. If locale is invalid, use project default.'