Single tool 'seo_research_keywords' has a well-structured input schema with proper type and description for the 'query' parameter. The tool description is comprehensive (380+ chars) and includes use cases, output structure, and anti-patterns (what NOT to use it for). However, the output schema is documented only in prose within the description rather than as a formal JSON Schema object. Parameter descriptions are present but minimal. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) are declared. The tool follows verb_noun naming ('seo_research_keywords') and accepts a natural-language parameter ('query' as a string seed keyword), which aligns with chat-data-model principles. Risk classification as READ_ONLY is correct but not formally encoded in the schema.

• Add formal JSON Schema definition for the output object with typed fields: suggestions (array of strings), relatedQueries (object with keys vs/for/howTo/etc mapping to arrays), longTailKeywords (array), searchIntent (enum: informational|transactional|navigational|commercial), totalKeywords (integer).
• Declare tool annotations in the schema: add 'readOnlyHint: true' to indicate the tool does not modify state and is safe to call multiple times.
• Add constraints to the 'query' parameter: minLength (e.g., 1), maxLength (e.g., 200), and a description note like 'Seed keyword or phrase (1-200 chars). Examples: bitcoin trading, best crm software.'
• Document error scenarios: 'Returns error if query is empty, contains only whitespace, or if the SEO data source is unavailable. Retry with a simpler or more specific keyword.'
• Consider adding optional parameters for result filtering: 'limit' (default 20, max 100) to cap suggestions/keywords returned, and 'intent_filter' (enum: informational|transactional|navigational|commercial) to narrow results by search intent.

Inferred effective spec: 2025-06-18+.

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