Three tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. write_artifact and read_artifact have minimal descriptions (under 100 chars). No parameter type constraints (enums, ranges). Output schemas not documented. No error handling guidance. Naming is verb-noun compliant but descriptions lack WHEN/WHY context. No security considerations documented despite write_artifact being a WRITE operation.

• Expand write_artifact description to 100-150 chars: 'Store text in the shared workspace under an object key. Use this to hand work to another agent instead of pasting it into your answer. Returns success status and the stored key. Overwrites existing content at that key without confirmation.'
• Document output schemas: write_artifact returns {success: bool, key: string, size_bytes: int}; read_artifact returns {content: string, key: string, size_bytes: int} or null if not found; list_artifacts returns {keys: [string], count: int, prefix: string}.
• Add parameter constraints: prefix max 256 chars, key must match ^[a-zA-Z0-9/_.-]+$, content max 10MB. State these in parameter descriptions.
• Add error handling: 'If key format is invalid, returns error with pattern requirement. If quota exceeded, returns error with current usage. If permission denied, returns error with required scopes.'
• Document idempotency: 'write_artifact is idempotent, calling twice with the same key and content produces the same result. Calling with different content overwrites without warning.'