Todoist

**Judgment:** A solid, task-focused Todoist integration that maps well to natural language task-management requests. It appears mature enough for everyday personal productivity use, especially for listing, adding, and updating tasks. **Strengths:** - Supports the essential Todoist operations: listing tasks (today/overdue/all), filtering by priority, project, or label, and searching by keyword. - Allows rich task creation (due dates, priority, project, labels, and insertion order) which an AI assistant can use to create structured tasks from a user’s natural language. - Includes project/label listing and creation, plus comments, which broadens it beyond bare-minimum task operations. **Key Risks / Limitations:** - **Destructive operations:** `delete`, `done`, `move`, and `update` are permanent or behavior-changing actions; if the AI misidentifies a task (e.g., due to ambiguous search results), it could complete or delete the wrong item. - **Ambiguity in task selection:** The documented pattern is `search` → use task ID → act. If multiple matches exist and disambiguation is not handled carefully in the agent logic, the wrong task may be modified. - **Natural-language due dates:** Todoist’s date parsing (e.g., “next Monday”, “tomorrow”) may not match user intent in edge cases (time zones, ambiguous phrases); the agent should confirm important or long-term deadlines. - **API token handling:** Requires a Todoist API token via `todoist auth` or environment variable; any misconfiguration or insecure handling is a security risk outside the skill’s direct control. **Recommended Scenarios:** - Everyday personal task and reminder management (today’s tasks, add a task with a due date, mark done, adjust due dates). - Project-focused productivity workflows where the assistant regularly queries and maintains tasks within specific projects (e.g., “Work”, “Personal”). - Priority-based triage sessions (e.g., “Show my high priority tasks”, “Move these to the Work project and set priority 1”). **Use With Extra Care:** - When performing bulk or destructive actions (deleting tasks, mass-completing tasks) or acting on tasks found via fuzzy/short search queries. - For critical deadlines or rare, important tasks where ambiguous natural-language dates could cause serious consequences; in such cases, the assistant should double-check details with the user before executing.