Two tools with explicit Zod schemas and descriptions, but significant gaps in parameter clarity and output documentation. fix_pr_comments has a clear, verb-led name and reasonable description (127 chars), but mark_comments_as_handled uses a compound verb ('mark...as') and lacks clarity on when it's safe to call. No output schema documented. Error handling present but generic. Schemas are visible and well-formed, but descriptions could be more LLM-optimized.

• Rename 'mark_comments_as_handled' to 'resolve_comments' or 'close_comments', single action verb, clearer intent for LLM parsing.
• Convert reaction examples to a formal enum in the Zod schema: z.enum(['+1', '-1', 'laugh', 'confused', 'heart', 'hooray', 'rocket', 'eyes']). Remove examples from description text.
• Document the output schema for both tools. Specify the structure of PR comments returned by fix_pr_comments (e.g., fields: id, author, body, status, created_at) and the response structure for mark_comments_as_handled (success count, failed IDs, etc.).
• Add maxLength and minLength constraints to fixSummary in the Zod schema (e.g., z.string().min(3).max(50)) instead of relying on prose descriptions.
• Add a 'dry_run' parameter to mark_comments_as_handled to support confirmation before executing. This prevents accidental bulk comment closures and aligns with the confirmation-request pattern.

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

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