The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes an external file, transcript.json, to generate social media post text. This creates a surface for indirect prompt injection where malicious instructions inside the transcript could influence the agent's output.
Ingestion points: The agent reads transcript.json in Step 5 to generate post content.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the transcript data.
Capability inventory: The skill utilizes several local tools including node render.mjs, video-sound, call-to-action, and video-check.
Sanitization: No sanitization or validation logic is defined for the content of the transcript.
[COMMAND_EXECUTION]: The skill instructs the agent to execute several local command-line tools as part of the video production workflow.
Evidence: The instructions call for running node render.mjs with various flags, as well as standalone commands like video-sound, call-to-action, and video-check.
Context: These executions appear to be the primary intended function of the skill for video rendering and quality assurance.

every-format