Loading ToolBench
Content-as-a-Service MCP Server — agent-to-agent article generation and NPC dialogue creation
Two tools with complete schemas and reasonable descriptions. Both tools have proper Zod schema definitions with typed parameters and enums. Descriptions are adequate (100-150 chars) but lack actionable context for LLM selection. Parameter descriptions are present but generic. Error handling returns isError flag but lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint/idempotentHint). Output schemas are not documented, responses are JSON stringified without field-level documentation.
Generate a complete, publish-ready article with matched images. Send a topic, get back a finished article.
Generate structured NPC dialogue with greetings, contextual responses, rumors, and idle chatter. Returns tagged JSON ready for game engines.
Output schemas not documented. Both tools return JSON-stringified results without field-level documentation. LLMs cannot plan downstream operations or extract specific fields reliably.
Error messages lack recovery guidance. Errors return isError=true with plain text but do not suggest next steps (e.g., 'Rate limit exceeded. Wait 60s or reduce word_count.').
Tool descriptions lack WHEN-to-use context. 'Generate a complete, publish-ready article' does not explain when to call this vs. other content tools or what prerequisites exist.
No tool annotations. Neither tool declares readOnlyHint, destructiveHint, or idempotentHint. Both are WRITE operations but lack explicit marking.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | <=2025-11-25 | v2 |
Parameter descriptions are minimal. 'The article topic' and 'NPC's name' lack format/constraint guidance. No hints on length limits, character restrictions, or dependencies.