A Next.js-based agent framework with skill management, markdown processing, and EPUB conversion capabilities
Both tools have critical definition gaps. markdown_to_epub has a reasonable description (89 chars) and partial schema (4 params with types but missing descriptions for 3 of them). idea_refine has a vague description (67 chars) and minimal schema (1 param, 'action' type string but no enum or valid values documented). Neither tool documents output schemas. No error handling guidance. Tool names lack clear verb-noun structure ('idea_refine' is ambiguous, refine what? how?). Parameters lack actionable constraints and format guidance. No evidence of idempotency, pagination, or composition patterns.
Initialize and manage the ideas directory for the idea-refine skill
Convert markdown files to EPUB3 format with proper chapter structure, styling, and table of contents
Tool names lack action verbs. 'markdown_to_epub' is descriptive but not verb-led; 'idea_refine' is vague and does not signal what action occurs. LLMs cannot infer intent from these names alone.
Parameters lack descriptions. markdown_to_epub has 4 params; only 2 have descriptions ('markdown_file', 'output_path'). 'title' and 'author' lack any guidance on format, length, or constraints. idea_refine's 'action' param has no description of valid values (e.g., is 'init' the only action? what do other actions do?).
No output schemas documented. Neither tool specifies what it returns. LLMs cannot plan downstream calls or extract results without knowing the response structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 20 | 2026-07-28+ | v2 |
idea_refine's 'action' parameter is unconstrained. No enum, no valid values listed, no format guidance. LLMs will hallucinate action names. Should declare: action must be one of: init, list, add, remove, etc.
No error handling guidance. Both tools are marked WRITE risk but provide no recovery hints. If markdown_to_epub fails (invalid markdown, disk full, permission denied), the LLM has no guidance on what to do next.