The Agent Skills Directory

[COMMAND_EXECUTION]: Uses curl to communicate with a local Task HTTP API at http://localhost:$PORT/api/tasks. This is the primary mechanism for creating, updating, and managing scheduled tasks.
[PROMPT_INJECTION]: The skill implements a feature to store and execute user-provided prompts at a later time, which introduces a surface for indirect prompt injection.
Ingestion points: User input is mapped to the prompt field in the API request body as documented in SKILL.md.
Boundary markers: Absent; the instructions do not prescribe delimiters or protective markers to wrap the scheduled prompt to prevent its contents from being interpreted as agent instructions during execution.
Capability inventory: The system can execute prompts and transmit results to various communication channels (e.g., Slack, Telegram, Discord).
Sanitization: No sanitization or content validation is performed on the stored prompt before it is sent to the API.

task-manager