Single tool 'execute_skill_action' has a vague, marketing-focused description (97 chars) that does not explain WHAT the tool does, WHEN to use it, or WHAT it returns. Parameter 'query_payload' is a generic string with minimal guidance, no format, constraints, or examples. No output schema documented. The tool name uses 'execute' (generic verb) + 'skill_action' (vague noun), violating verb_noun clarity. No error handling, no pagination, no structured output guidance. Implementation echoes input without demonstrating actual CRM deduplication logic from client.py. This server reads like a stub or demo, not production-ready.

• Rename tool to a specific action verb + noun pair. If it deduplicates CRM records, call it 'deduplicate_crm_records' or 'merge_contact_records'. If it enriches leads, call it 'enrich_lead_data'. The name alone should convey intent.
• Replace description with functional documentation: 'Identifies duplicate CRM records (by email, domain, or fuzzy company name match) and merges them using source-weight-based conflict resolution. Returns merged record and audit log of field-level decisions.' (This matches the client.py logic.)
• Add parameter constraints: 'query_payload (string, required): JSON object with keys: email (string), domain (string), company_name (string), source (enum: salesforce_master|direct_user_form|apollo_enrichment|web_scraper|legacy_import). Example: {"email": "[email protected]", "company_name": "Acme Inc", "source": "apollo_enrichment"}'
• Document output schema: 'Returns object with action (enum: MERGE_UPDATE|CREATE_NEW), matched_record_id (string|null), match_reason (string), merged_payload (object), resolution_audit (array of {field, action, old, new} objects).'
• Add error handling: validate query_payload is valid JSON; return 'Invalid query_payload: must be valid JSON object with email, domain, company_name, source fields' if parsing fails. Classify errors as user-fixable (invalid JSON, missing required fields) vs fatal (internal service error).