Single tool 'convert_to_markdown' has a reasonable description (89 chars) but lacks critical schema documentation. Input schema shows only 'uri' parameter with basic type and description, but no constraints (format, pattern, length limits), no output schema documentation, and no error handling guidance. Tool name is clear (verb_noun pattern), but parameter description is minimal. No parameter validation rules, no enum constraints for supported formats, no pagination/limits, and no structured output schema visible. Follows basic naming convention but falls short on schema completeness and LLM-optimized documentation.

• Add explicit output schema documenting the markdown content structure (e.g., { 'markdown': string, 'format': string, 'pages': number, 'metadata': object }). This enables LLMs to extract and chain results.
• Constrain 'uri' parameter with format='uri' and pattern validation. Document accepted schemes: 'file:// for local paths, http(s):// for remote URLs. Specify max length (e.g., 2048 chars).'
• Replace format list in description with an enum constraint: ['pdf', 'docx', 'pptx', 'xlsx', 'png', 'jpg', 'gif', 'webp', 'svg']. This prevents hallucinated formats.
• Add error handling section to description: 'Returns error if file not found, format unsupported, or conversion fails. Errors include: file_not_found, unsupported_format, conversion_timeout, invalid_uri.'
• Expand parameter description to ~100 chars: 'File URI (file:// for local paths, http(s):// for remote) or absolute local path. Max 2048 chars. Relative paths not supported.'