10 tools with complete input schemas and descriptions, but quality is uneven. Tool names follow verb_noun convention (init_, resume, next_, diff_, verify_, mark_, skip_, get_). Descriptions are present but brief (10-50 chars), lacking context on WHEN to use each tool or what happens to state. Parameters have types but many lack descriptions (e.g., 'configPath' in init_migration and resume has no description text visible in schema). Output schemas are not documented, LLMs cannot predict what fields to expect. Error handling is minimal; no recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Risk labels (READ_ONLY, WRITE) are metadata but not exposed in MCP schema.

• Add output schema documentation to every tool. E.g., 'Returns: { discovered: number, auth: { type: string }, estimatedSeconds: number }' for init_migration.
• Expand tool descriptions to 50-150 chars. Include WHEN to use, prerequisites, and state side effects. E.g., 'Initialize migration from config. Discovers pages via sitemap/crawl. Returns page count and estimated runtime. Call this first before any other tool.'
• Add descriptions to all parameters. E.g., configPath: 'Absolute or relative path to page-to-page.config.json. File must exist and be valid JSON.'
• Add tool annotations to MCP schema. Mark read-only tools with readOnlyHint: true; destructive tools (mark_matched, skip_current) with destructiveHint: true.
• Implement error recovery guidance. E.g., if config not found: 'Config file not found at {path}. Ensure page-to-page.config.json exists in the project root. See https://github.com/alexeymoskalev-devops/page_to_page#configuration for setup.'