Server has 30 tools with consistent Zod schemas and descriptions. Naming follows verb_noun pattern (getGanttData, createIssue, updateIssue). However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool vs. similar ones. Parameter descriptions exist but are minimal (avg ~40 chars). Output schemas are not documented, callers must infer structure from implementation. Error handling is basic (try/catch with formatApiError). No tool annotations (readOnlyHint, destructiveHint) despite clear risk levels. Missing idempotency guidance for write operations.

• Add output schema documentation to every tool. Document the structure of returned objects, field types, and which fields are safe to chain into downstream tools (e.g., createIssue returns issueId, which updateIssue accepts).
• Expand tool descriptions to 100-150 chars. Include WHEN to use the tool: 'Get all issues in a project (use getAllIssues for bulk retrieval; use searchIssues for filtered queries; use getMyIssues for current user only).'
• Add tool annotations to the MCP tool registration. Mark deleteIssue and deleteIssueLink with destructiveHint=true; mark read-only tools with readOnlyHint=true; mark idempotent operations with idempotentHint=true.
• Enhance error messages with recovery guidance. Instead of 'Failed to get Gantt chart data: <error>', return 'Failed to get Gantt chart data: <error>. If the project does not exist, call listProjects() to find valid project IDs.'
• Add numeric bounds and format constraints to parameter descriptions. E.g., 'estimationMinutes (1 - 480 minutes)', 'timezone (IANA timezone string, e.g., America/New_York)', 'limit (1 - 100, default 20)'.