Two tools with basic schemas and descriptions, but significant gaps in parameter documentation, output schema clarity, and error handling guidance. fetch_ticket lacks parameters for include_comments/include_attachments in the schema (only hardcoded to true in implementation). search_tickets has minimal parameter descriptions. No output schemas documented. Error handling returns plain text strings without recovery guidance. Descriptions are adequate (100-150 chars) but lack WHEN/WHY context. No tool annotations (readOnlyHint, idempotentHint). Naming is clear (verb_noun pattern) but parameters lack type hints and constraints.

• Add include_comments and include_attachments as optional boolean parameters to fetch_ticket schema. Update description to explain when each is useful (e.g., 'include_comments: Fetch all comments on the ticket. Useful for understanding discussion history.').
• Document output schemas for both tools. Define the structure of returned JSON: fetch_ticket returns {key, summary, description, status, assignee, reporter, created, updated, priority, labels, components, custom_fields, comments?, attachments?}. search_tickets returns [{key, summary, status, assignee, priority}].
• Replace generic error strings with structured error responses. Example: 'Ticket PROJ-999 not found. Try search_tickets(jql="key ~ PROJ") to find similar tickets.' Categorize as retryable (network timeout) vs user-fixable (invalid key).
• Add constraints to search_tickets maxResults: 'Maximum number of results to return (1 - 100, default 50). Larger values may timeout or exhaust API quota.'
• Add readOnlyHint=true to both tool definitions in list_tools(). Update tool descriptions to explicitly state 'This tool does not modify Jira data.' This signals to LLMs that calls are safe to retry.