The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the twelvelabs-pp-cli binary using npx or go install from repositories associated with the vendor mvanhorn. These are documented as part of the tool's setup process.
[DATA_EXFILTRATION]: The CLI includes a --deliver webhook:<url> flag. This feature allows the output of any command to be sent to a remote URL via an HTTP POST request, which could potentially be used to exfiltrate video metadata or other command results if an agent is misdirected.
[COMMAND_EXECUTION]: The skill relies on the execution of the twelvelabs-pp-cli binary and ffmpeg to perform video processing tasks like uploading, indexing, and clipping.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured video data (summaries, highlights, and chapters) retrieved from an external API. This establishes an attack surface where malicious text embedded in video content could influence subsequent agent decisions.
Ingestion points: Data is ingested via the video-brief and upload-video commands in SKILL.md.
Boundary markers: The instructions do not specify any delimiters or warnings to ignore instructions embedded in the video data.
Capability inventory: The agent can execute CLI commands, write files, and perform network requests through the tool's webhook feature.
Sanitization: There is no evidence of sanitization or filtering of the content received from the Twelve Labs API.

pp-twelvelabs