Single tool 'search_records' has a minimal schema with only one parameter (resource_type: string). Description is present but generic (23 chars: 'Search for records in Attio by resource type'). No output schema documented. Parameter lacks format constraints, enums, or usage examples. No error handling guidance. Tool definition appears inferred from setup.sh rather than explicit MCP registration visible in source. Critical gaps in schema completeness, parameter documentation, and output structure.

• Add explicit enum constraint to resource_type parameter: e.g., enum: ['companies', 'contacts', 'deals', 'people'] with description 'The Attio resource type to search. Valid values: companies, contacts, deals, people.'
• Expand tool description to 100 - 150 chars: 'Search for records in Attio by resource type (companies, contacts, deals, people). Returns matching records with IDs, names, and metadata. Requires ATTIO_ACCESS_TOKEN and ATTIO_WORKSPACE_ID.'
• Document output schema explicitly: 'Returns { records: [{id, name, type, ...}], total_count, next_cursor }' so LLMs know what fields to expect and can chain to downstream tools.
• Add pagination parameters: limit (1 - 100, default 20) and cursor (for next_cursor from previous response) to prevent context window exhaustion.
• Include error recovery guidance in description: 'If resource_type is invalid, returns error with list of valid types. If no results found, returns empty records array.'