The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests external video clips for analysis and comparison, creating a potential surface for indirect prompt injection where instructions hidden in video metadata, audio, or frames could influence agent behavior.
Ingestion points: Reference video clips provided for analysis in Step 1 and Step 6.
Boundary markers: Absent. The skill does not define specific boundaries or instructions for the LLM to ignore embedded content within the analyzed clips.
Capability inventory: The skill has access to file system operations (agent-fs write), video rendering tools (remotion), and general command execution (ffmpeg).
Sanitization: No sanitization or filtering of input video metadata or content is described.
[COMMAND_EXECUTION]: The skill uses various command-line utilities for its primary workflow, including ffmpeg for complex filter operations and frame extraction, sha256sum for verifying the integrity of brand assets, and the remotion CLI for rendering video compositions. It also executes a script named fetch-music.sh to download audio tracks.
[EXTERNAL_DOWNLOADS]: The skill references and downloads brand assets from external sources, specifically https://agent-swarm.dev/logo.png. This domain is recognized as part of the vendor's official project infrastructure and the download includes an integrity check using sha256sum.

motion-design-replication