All 8 tools have basic descriptions (50-100 chars) and verb-noun naming (get*Diagram). However, input schemas are severely underdeveloped: each tool accepts a single complex object parameter (ClassDiagram, ComplexDiagram, ErDiagram, etc.) with NO visible type definitions, field descriptions, or constraints in the source code. The parameter descriptions are minimal ('The X diagram object to convert to mermaid format'). No output schemas are documented. No error handling guidance. No parameter validation rules. Tools lack the structured, LLM-optimized definitions required for reliable agent composition. Naming is consistent but generic ('get' prefix repeated 8 times with minimal differentiation). No pagination, no batch operations, no idempotent hints.
Generate a complex diagram in D2Lang declarative grammar; for example software architecture, network architecture, any diagram that contains containers or grouping. An URL will be returned to open the diagram, you have to show this URL
Generate a markdown mermaid class diagram. An URL will be returned to open the diagram, you have to show this URL
Generate a markdown mermaid complex diagram; for example software architecture, network architecture, any diagram that contains containers or grouping. An URL will be returned to open the diagram, you have to show this URL
Generate a markdown mermaid Entity Relationship diagram. An URL will be returned to open the diagram, you have to show this URL
Generate a markdown mermaid gantt diagram. An URL will be returned to open the diagram, you have to show this URL
Generate a markdown mermaid mind map. An URL will be returned to open the diagram, you have to show this URL
No input parameter schemas visible. Each tool accepts a complex object (ClassDiagram, ComplexDiagram, etc.) but no JSON Schema with field types, required fields, or constraints is documented in source code. Per HARD SCORING RULE: schema score MUST be 0 when no schema is visible.
Parameter descriptions are minimal and generic. 'The X diagram object to convert to mermaid format' does not explain what fields the diagram object requires, what constraints apply, or what happens if invalid data is passed. LLMs cannot infer structure from generic descriptions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
Generate a markdown mermaid sequence diagram. An URL will be returned to open the diagram, you have to show this URL
Generate a markdown mermaid timeline. An URL will be returned to open the diagram, you have to show this URL
No output schema documented. Tools return a string URL ('Diagram is available on : https://...') but LLMs have no formal schema describing the response structure, field names, or how to extract and use the URL.
No error handling or recovery guidance. Tools throw IOException but provide no guidance on what errors are possible, whether they are retryable, or what the LLM should do next. Per pattern:recovery-guide, errors must tell the agent what to do.
Tool names lack specificity. Eight tools all start with 'get' and differ only in diagram type. Names like 'getMermaidClassDiagram' vs 'getMermaidComplexDiagram' do not clearly convey when to use each. Per naming baseline, 90% of A+ tools start with action verbs, but verb alone is insufficient, the noun must disambiguate.