The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were detected in the skill's instructions or implementation scripts. The skill follows best practices for tool development within the Zapier ecosystem.\n- [DATA_EXPOSURE]: The skill interacts exclusively with the Microsoft Graph API (graph.microsoft.com) to manage To Do resources. All network operations are performed using the SDK-provided fetch wrapper, and the skill does not attempt to access sensitive local files such as SSH keys or AWS credentials.\n- [CREDENTIALS_UNSAFE]: Authentication is handled via Zapier-managed connections or user-provided environment variables (e.g., MICROSOFT_TODO_ACCESS_TOKEN). The documentation explicitly guides users to manage these secrets securely using environment variables or secret stores rather than pasting them in plain text.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect injection surface as it ingests untrusted data from task titles and notes (ingestion points: listTasks.ts, findTask.ts, getTask.ts). However, it implements robust boundary markers and sanitization through Zod output schemas (taskSchema, checklistItemSchema) in lib/microsoft-todo.ts which strip unknown fields. Additionally, the skill escapes OData queries using the odataQuote helper in scripts/findTask.ts.\n- [COMMAND_EXECUTION]: The cli.js entry point uses spawnSync to delegate execution to the connector's internal logic. This implementation is used for runtime compatibility (e.g., handling differences between Node.js and Bun) and does not provide a path for arbitrary shell command execution from user input.

microsoft-todo