hSUM has 4 tools with basic descriptions and partial schemas. Tool names follow verb_noun convention (evidence_search, evidence_get, evidence_project, evidence_status), which is good. However, descriptions are minimal (10-50 chars), parameter descriptions are missing entirely, and input schemas lack type information beyond string/boolean. The evidence_project tool conflates multiple actions (list, create, delete) into a single tool with an 'action' enum, violating single-responsibility. No output schemas are documented. Error handling is not visible in the provided code. Tool definitions appear inferred from shell scripts rather than explicit MCP registration code.

• Locate and review the actual MCP server implementation (likely in src/runtime.rs or similar). Verify tool registration with explicit schema definitions, not inferred from shell scripts.
• Expand all tool descriptions to 50-200 characters. Example: 'evidence_search' → 'Search indexed evidence sources using full-text query. Returns matching documents with citation URIs and relevance scores. Use when you need to find relevant evidence for a claim.'
• Add descriptions to every parameter. Example: 'query' → 'Full-text search string. Supports AND, OR, phrase queries. Max 1000 chars.'
• Split evidence_project into separate tools: evidence_project_list (no params), evidence_project_create (project_name), evidence_project_delete (project_name). Each tool should do one thing.
• Document output schemas for all tools. Example: evidence_search returns {results: [{citation_uri: string, title: string, relevance: number}], total_count: number}.