Bibverify provides 4 well-defined tools with clear verb-based names and reasonable descriptions. All tools have documented input schemas with type information and parameter descriptions. However, several gaps limit quality: (1) output schemas are not explicitly documented in the source, forcing LLMs to infer response structure; (2) parameter descriptions are functional but minimal (averaging ~30-50 chars), lacking actionable constraints and format guidance; (3) no error handling guidance beyond ToolError exceptions; (4) descriptions are adequate (60-90 chars each) but lack strategic details about when/why to use each tool; (5) no tool annotations (readOnlyHint) despite all tools being read-only operations. The tools are well-composed (each does one thing) and follow appropriate naming conventions. Parameter typing is present but sparse.

• Document output schemas for all tools in mcp_server.py. Add a docstring or type hint showing the structure of the response (e.g., 'Returns: {bibtex_entry: str, doi: str, metadata: {...}}'). This enables LLMs to plan multi-tool workflows.
• Expand tool descriptions to 100-150 chars, explaining not just what the tool does but when an LLM should choose it. Example for doi_to_bibtex: 'Fetch a BibTeX entry from Crossref using a DOI. Use this to enrich incomplete references or verify citation metadata before bulk verification.'
• Add format and constraint guidance to parameter descriptions. E.g., 'The DOI to fetch (standard format: 10.xxxx/yyyy, case-insensitive).' For optional parameters, state the default: 'Optional BibTeX entry key. If omitted, a key is auto-generated from author and year.'
• Add tool annotations to schema definitions. Set readOnlyHint=true for all 4 tools to signal they are safe read-only operations. Example: tools[...].set_annotation('readOnlyHint', True).
• Enhance error messages with recovery guidance. When workspace validation fails, include: 'ToolError: config path outside workspace. Ensure all paths in config.json are relative to workspace_root or check that workspace_root is set correctly.'
• Add parameter dependency documentation. For config_file, clarify: 'If provided, overrides default config.json. Must be relative to workspace_root or an absolute path within it.'
• Consider adding a discover/list_config_values tool to help LLMs understand available providers and settings without trial-and-error.