Two tools with complete input/output schemas and clear descriptions. Tool names follow verb_noun pattern (pdf.read, pdf.info). Descriptions are adequate (100-150 chars) but lack actionable context for LLM selection. Parameters are well-typed with descriptions, but lack constraint documentation (e.g., page range format, maxPages/maxChars rationale). Output schemas are defined but not visible in source excerpt. Error handling and recovery guidance absent. No tool annotations (readOnlyHint, idempotentHint) despite both being read-only operations.

• Add readOnlyHint: true to both tool definitions to signal safety to agents.
• Expand descriptions to 150-250 chars with WHEN/WHY context: 'Use pdf.read to extract text content; use pdf.info to check page count and encryption status. Set structured=true for per-item positions (slower, larger output).'
• Document parameter constraints in descriptions: pages format ('1-5' or '1,3,5'), maxPages rationale ('default 50 to prevent memory exhaustion'), maxChars rationale ('default 200k to fit in context window').
• Implement structured error responses: return {error: string, code: string, suggestion?: string} for file not found, encryption, permission, and timeout errors. Guide LLM to retry with password or check path.
• Verify output schemas include all fields needed for agent reasoning: for pdf.read return {text, pages_extracted, total_pages, char_count, truncated?}; for pdf.info return {pages, encrypted, permissions, metadata}.