todoist latest

**Quick judgment**: This skill provides a strong integration with Todoist via a mature CLI (`todoist-cli-rs` / `td`), with broad support for listing, adding, editing, and completing tasks, plus project/label management. It is well-suited for power users who already rely on Todoist and are comfortable with a CLI-backed workflow. **Key strengths** - **High functional coverage** of Todoist features, including natural-language quick add, powerful filters, and project/label management. - **Good performance** via local caching for reads and explicit sync commands for up-to-date data when needed. - **Workflow-oriented design**, with explicit guidance for morning review, quick capture, focus lists, and end-of-day checks. **Main risks / limitations** - **Environment dependency**: The skill requires `td` to be installed and configured (API token, initial sync). If the binary is missing or misconfigured, calls will fail. - **Cache staleness**: Reads default to cached data; if the agent or user forgets to use `--sync`/`td sync`, task views may be outdated relative to the live Todoist account. - **Destructive operations**: Commands like `td delete` and `td done --all-occurrences` can permanently alter or end recurring tasks; the agent should ideally confirm intent with the user before issuing such commands. - **Parsing/output considerations**: The CLI output format is not documented here; reliably parsing IDs and task details may require some care in prompt design or command selection. **Recommended scenarios** - Users who explicitly say they use **Todoist** and want to manage “my tasks”, “today’s tasks”, or “task list” directly from the assistant. - Chat-based **task capture and review** flows (e.g., “add a task for tomorrow…”, “show my high-priority work tasks today”). - Power users who want **filter-based views** (e.g., next-actions, waiting-on, overdue + high-priority) driven by natural-language requests. **When to be cautious or avoid** - Environments where the `td` CLI is not installed or cannot be installed (restricted/locked-down systems). - Users who are not actually Todoist users or are unsure whether their Todoist is connected/configured. - Automated flows that might aggressively complete or delete tasks without explicit user confirmation.