Critical gaps across all dimensions. Tool definitions are inferred from file paths and brief descriptions only, no visible input schemas, parameter types, or output documentation in provided source. Descriptions are extremely terse (10-40 chars), well below the 194-char baseline for production tools. No evidence of parameter validation, error handling guidance, or schema enforcement. The 6 tools lack the structured definitions required for reliable LLM invocation.

• Create explicit tool registration with full JSON Schema input definitions. Each parameter must have type, description (50 - 100 chars), and constraints (enum, min/max, pattern). Example: { name: 'dispatch_scout', description: 'Run reconnaissance on a target domain. Returns DNS records, WHOIS data, and SSL certificate info. Use this first to gather baseline intelligence before profiling.', inputSchema: { type: 'object', properties: { domain: { type: 'string', description: 'Target domain (e.g., example.com). Must be a valid FQDN.' }, depth: { type: 'integer', enum: [1, 2, 3], description: 'Reconnaissance depth: 1=DNS only, 2=DNS+WHOIS, 3=DNS+WHOIS+SSL.' } }, required: ['domain'] } }
• Expand tool descriptions to 150 - 250 characters. Include: (1) what the tool does, (2) when to use it vs similar tools, (3) what it returns. Example: 'Dispatch Scout performs passive DNS reconnaissance on a target domain. Returns A/MX/TXT records, WHOIS registrant info, and SSL certificate details. Call this first to establish baseline intelligence before running Profiler for deeper OSINT.'
• Add error handling and recovery guidance. For WRITE tools (dispatch-scout, dispatch-harvester, dispatch-profiler, write-outreach), implement dry-run mode or confirmation step. Return errors like: 'Domain lookup failed: example.com is not a valid FQDN. Ensure domain has at least one dot and valid TLD.' instead of raw HTTP 400.
• Rename tools to use underscores: dispatch_scout, dispatch_harvester, dispatch_profiler, write_outreach, get_job_status, get_signals. This matches MCP convention and improves LLM parsing.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry