Lune has well-structured tool definitions with consistent naming (verb_noun pattern), clear descriptions (avg ~120 chars), and complete input schemas. All 15 tools follow the search/get/list/extract pattern. However, output schemas are not explicitly documented in the visible code, parameter descriptions lack constraint details (ranges, formats, enums), and error handling guidance is minimal. No tool annotations (readOnlyHint) despite all being read-only. Descriptions are adequate but could be more prescriptive about when to use each tool vs. alternatives.

• Add explicit outputSchema to every tool definition. Document field names, types, and pagination structure (e.g., 'Returns {papers: [{id, title, authors, year}], total_count, next_cursor}').
• Enhance parameter descriptions with constraints: 'query (1-500 chars, natural language preferred)' for search tools; 'paper_id (arXiv ID or DOI format)' for ID params; 'limit (1-100, default 20)' for pagination.
• Add readOnlyHint: true to all 15 tool definitions in the schema to signal safety and enable agent optimizations.
• Implement structured error responses with recovery hints: 'Paper not found. Try search_papers() with keywords from the title, or verify the paper_id format (arXiv or DOI).'
• Add 'when to use' guidance to tool descriptions: 'Use search_papers for broad discovery. Use search_related_papers after finding a key paper to explore the citation graph. Use search_papers_many for systematic literature sweeps across multiple angles.'

Inferred effective spec: 2026-07-28+.

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