Single tool with comprehensive schema and detailed descriptions. Tool naming is clear (verb_noun pattern). Parameter descriptions are thorough with constraints, ranges, and format guidance. Output schema is documented in description. Error handling is present but could be more explicit about recovery paths. No security issues detected (API key via env var). Main gaps: no tool annotations (readOnlyHint/destructiveHint), output structure not formally documented in schema, and error categorization could guide LLM recovery better.

• Add formal output schema documentation: define return object with fields (success: boolean, output_path: string, document_id?: string, document_url?: string, document_expires?: string, file_size_bytes?: number). Include in tool description or as a separate schema block.
• Add tool annotations: mark as destructiveHint=false (read-only file write), idempotentHint=true (same input → same PDF), and readOnlyHint=false (modifies filesystem).
• Enhance error messages with recovery guidance. E.g., 'API key missing or invalid. Set PDFPIPE_API_KEY environment variable.' or 'Timeout after 60s. Try reducing page complexity or increasing timeout_ms parameter.'
• Validate mutually exclusive parameters at schema level using Zod discriminated unions or explicit validation that returns a clear error if both html and url are provided.
• Document parameter dependencies: 'filename is only used when store=true. If store=false, filename is ignored.' Add runtime validation to warn if filename is passed without store=true.