Fetchira provides 6 tools with basic naming and descriptions, but critical gaps undermine production readiness. Tool names follow verb_noun convention (search, read, browser, create_image, deep_research, usage), which is good. However, descriptions are inconsistent in depth and actionability. Input schemas are present but lack comprehensive parameter descriptions and type constraints. The 'browser' tool accepts vague action strings without enum constraints. Output schemas are not documented. Error handling guidance is absent, tools do not indicate what to do on failure or which errors are retryable. No tool annotations (readOnlyHint, idempotentHint) despite all being read-only or side-effect-free. Parameter descriptions are minimal (e.g., 'Optional specific provider to use' lacks guidance on valid provider names). The 'usage' tool has empty input schema with no description of what it returns.

• Add enum constraint to browser.action parameter: ['screenshot', 'click', 'type', 'scroll', 'wait', 'navigate']. Document what each action returns.
• Document output schema for each tool. E.g., search returns {results: [{title, url, snippet, provider}], total_count, next_cursor}. Include all fields the LLM needs for chaining.
• Expand parameter descriptions with valid values and constraints. E.g., provider: 'One of: serper, tavily, exa, firecrawl, grok. Defaults to quota-aware routing.'
• Add tool annotations: mark all tools with readOnlyHint=true since they do not modify state. Mark idempotent tools (search, read, usage) with idempotentHint=true.
• Document usage tool output schema and add description: 'Returns current quota usage and remaining limits across all configured provider accounts. Useful before calling search/read to check availability.'

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

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