The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted input to generate public social media content, creating a surface for indirect prompt injection.\n
Ingestion points: User-provided strings are processed via the prompt parameter in /api/generate and the topic parameter in /api/week (SKILL.md).\n
Boundary markers: The skill explicitly includes a 'Confirm Before Publishing' protocol, instructing the agent to display the exact native captions to the user and obtain explicit consent before performing any write operations.\n
Capability inventory: The skill uses network operations (curl) to interact with the PostWire API, which has the capability to publish content to external social media platforms.\n
Sanitization: No automated sanitization of the input prompt is described; the skill relies on the 'Confirm Before Publishing' step as a manual verification barrier.\n- [COMMAND_EXECUTION]: The skill instructions include multiple examples using the curl command-line tool to interact with the PostWire API.\n- [EXTERNAL_DOWNLOADS]: The skill enables the PostWire service to fetch media files from external URLs provided in the video_url and photo_url parameters.

postwire