ResearchForge has 8 tools with basic schemas and descriptions, but significant gaps prevent production use. All tools have descriptions (10-100 chars, mostly adequate), and input schemas are present with type definitions. However, output schemas are entirely undocumented, LLMs cannot plan downstream calls or extract structured data. Tool naming is inconsistent: mcp_read_resource and mcp_get_prompt use prefixes that obscure intent, while search_papers and retrieve_context lack clarity on scope. Parameters lack constraints (enums, ranges, format specs). Error handling is minimal, tools return plain strings with no recovery guidance. No tool declares permissions, audit trails, or idempotency guarantees. The retrieve_context tool has a Chinese description mixed with English, creating ambiguity. Overall, definitions are skeletal and would not pass code review.

• Document output schemas for all tools. Example: search_papers should return {papers: [{title: string, year: int, doi: string, summary: string}], total: int}. This enables LLMs to extract and chain results.
• Rename mcp_read_resource → read_resource and mcp_get_prompt → get_prompt. Drop the 'mcp_' prefix; context is clear from the registry parameter.
• Rewrite retrieve_context description in English only. Clarify scope: 'Search local knowledge-base documents (project docs, code notes, research notes, FAQs). Returns top-k matching chunks with source filenames.'
• Add parameter constraints: search_papers query should have max length 256; retrieve_context k should be 1 - 20 with default 4. Use enums where applicable (e.g., if paper types are fixed).
• Enhance error messages with recovery guidance. Example: 'Paper not found: "Quantum Computing". Try search_papers() with a partial title or author name.' This guides LLM retry logic.