The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses npx to run @useclawlink/cli for authentication and interacting with the Google Tasks API. This is the primary method for the skill's functionality.
[EXTERNAL_DOWNLOADS]: Fetches the @useclawlink/cli package from the public npm registry. As the package is owned by the skill's author (clawlink-hq), this is a standard and expected dependency.
[DATA_EXFILTRATION]: Stores authentication credentials in a local file at ~/.clawlink/credentials.json. This follows standard security practices for CLI tools to manage session keys without hardcoding secrets.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from task titles and notes.
Ingestion points: Data is pulled from external sources via googletasks_list_tasks, googletasks_list_all_tasks, and googletasks_get_task in SKILL.md.
Boundary markers: The instructions explicitly mandate that the agent must "confirm with the user before any write" and "Always require explicit user confirmation before invoking" destructive actions like clearing tasks.
Capability inventory: The skill has the ability to write, delete, and batch execute API operations through the CLI.
Sanitization: Relies on the CLI tool to properly handle and escape data before transmitting it to the Google Tasks API.

google-tasks