Model Context Protocol server for Wordware integration with dynamic tool loading from Wordware API
Server exhibits critical definition quality gaps. Tool schemas are present but descriptions are minimal (most under 50 chars). Parameter descriptions exist but lack detail on constraints, formats, and valid ranges. No output schemas documented. Error handling is absent, no recovery guidance or categorization. Tools lack verb-noun naming consistency (e.g., 'format_person_research' is a utility, not a primary action). The 'format_*' tools appear to be internal utilities exposed as public tools, violating single-responsibility principle. No evidence of idempotency, pagination support, or chaining-ID inclusion in responses.
Format the result of a Notion save operation.
Format person research results into readable text.
Format topic research results into readable text.
Research information about a person, such as a founder or executive.
Research a general topic using Wordware's research capabilities.
Save content to Notion using Wordware's integration.
Three 'format_*' tools expose internal formatting logic as public tools. These violate single-responsibility principle, they should be internal utilities, not agent-callable tools. Agents should not need to manually format API responses.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., does research_person return a person_id for chaining?). Violates pattern:tool requirement.
Parameter descriptions lack constraint details. 'full_name' has no guidance on format (first+last? aliases?). 'query' has no length limits or example topics. 'title' and 'body' for Notion have no length/format constraints. Violates pattern:constrained-input.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | <=2025-11-25 | v2 |
No error handling or recovery guidance. Code shows no try-catch blocks returning actionable errors. If Wordware API fails, agent receives raw exception, cannot retry or self-correct. Violates pattern:recovery-guide.
Tool names lack action verbs. 'format_person_research' and 'format_topic_research' are passive/utility names. Primary tools should be 'research_person' and 'research_topic' (which exist), but exposing formatters as separate tools confuses intent. Violates pattern:tool naming guidance.