MCP Server to interact with Asana API, providing tools to fetch tasks.
Single tool with basic but incomplete definitions. Tool naming follows verb_noun pattern (get_my_tasks_due_today) which is good, but the description is generic and lacks context for LLM selection. Input schema is empty (no parameters), which is correct but trivial. Output schema is present with type definitions but lacks field-level descriptions. No error handling guidance, no pagination support for results, and no actionable recovery instructions. The server exposes Asana credentials via environment variables correctly (not as parameters), which is a security win. However, the tool is READ_ONLY and very narrow in scope, only one action available limits practical utility.
Retrieves your Asana tasks that are due today from the configured workspace.
Description lacks selection context and prerequisites. 'Retrieves your Asana tasks that are due today from the configured workspace' does not explain when/why to call this vs. other task tools, whether workspace must be pre-configured, or what 'configured workspace' means operationally.
Output schema fields (tasksFound, summary) lack descriptions. The schema defines structure but does not explain what summary contains or how tasksFound relates to the actual task list returned. LLMs cannot infer whether summary is free text, structured JSON, or a status message.
No error handling or recovery guidance. Code returns {tasksFound: 0, summary: '...'} on failures (missing workspace GID, parse errors, API errors) but provides no actionable next steps. LLM receives 'Error: Received unexpected task data format' with zero guidance on what to do.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 53 | 2025-06-18+ | v2 |
| 2026-03-09 | F | 39 | - | v1 |
No pagination or result limiting documented. Code fetches tasks without documented limits, and the output schema only reports a count, not the actual tasks. For a 'due today' query this may be fine, but the response design does not follow pagination best practices (no next_cursor, no limit parameter for future expansion).
Tool is read-only and standalone. No complementary tools to create, update, or close tasks. Single-tool servers severely limit agent autonomy, agents can only query, never act. This is a missing capability rather than a definition flaw, but limits practical usage.
Output structure returns only summary and count, not the actual tasks. The code validates tasks against AsanaTaskSchema and logs them, but does not return the task array in the output. LLM receives a count but cannot see what tasks are due, defeating the purpose of the tool.