jiraACP has well-structured Zod schemas with type definitions and descriptions for all 6 tools. Tool names follow verb_noun convention (getTasks, getTicket, transitionTicket, addComment, reassign). However, descriptions are inconsistent in depth and some parameter descriptions lack actionable detail. Output schemas are not documented, LLMs cannot predict response structure. Error handling is minimal; most tools return raw JSON without recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear read/write distinction. Parameter descriptions sometimes reference examples (e.g. 'e.g. "hi", "geo"') which LLMs may reuse literally.

• Add explicit output schema documentation for each tool. E.g., getTasks returns {total: number, issues: [{key: string, summary: string, status: string, assignee: string, priority: string}]}. Include this in tool descriptions or as a separate schema field.
• Add tool annotations: mark getTasks, getTicket, getTransitions with readOnlyHint=true; mark transitionTicket, addComment, reassign with destructiveHint=true. This enables agents to reason about side effects.
• Replace example values with enum constraints. Change 'Instance name (e.g. "hi", "geo")' to an enum: {type: 'string', enum: ['hi', 'geo']} or document valid instances in the description without examples.
• Enhance error messages with recovery guidance. E.g., if reassign fails with 'Invalid accountId', return 'Invalid accountId: {value}. Call getTasks() first to find valid assignees.' This guides the agent to self-correct.
• Expand getTransitions description to explain its role: 'Discover available state transitions before calling transitionTicket. Returns transition IDs and names needed for state changes.' This clarifies the discovery pattern.

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

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