Latebra has 8 tools with explicit schemas and descriptions visible in server.py. Naming follows verb_noun convention (latebra_scrape, latebra_search, latebra_crawl). However, descriptions are generic and lack LLM-optimization guidance. Most parameters lack descriptions (e.g., latebra_check_anonymity's 'url' param has a description, but latebra_interact's 'selector' and 'text' lack context on format/constraints). Output schemas are not documented. Error handling and recovery guidance are absent. Tool latebra_interact is marked WRITE but has no confirmation/dry-run pattern. No tool annotations (readOnlyHint/destructiveHint) present despite MCP spec support.

• Add output schema documentation to each tool. Example for latebra_search: 'Returns {results: [{title: string, url: string, snippet: string}], total_count: integer, next_cursor?: string}'.
• Expand latebra_interact parameter descriptions: 'selector' → 'CSS selector string (e.g., #submit-btn, .form-input). Must match exactly one element.'; 'text' → 'Text to type. Special keys: {ENTER}, {TAB}, {ESCAPE}. Max 1000 chars.'
• Add tool annotations to inputSchema: latebra_scrape, latebra_search, latebra_crawl, latebra_check_anonymity → readOnlyHint: true; latebra_interact → destructiveHint: true; latebra_batch_scrape → idempotentHint: true.
• Implement error classification in tool responses. Return {success: false, error: string, category: 'retryable'|'user_fixable'|'fatal', suggestion: string} for failures.
• Add dry-run parameter to latebra_interact: 'dry_run?: boolean (default false). If true, returns what would happen without executing.' Prevents accidental clicks/typing.

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry