The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs extensive local command execution for media processing. It utilizes tools such as ffmpeg, ffprobe, yt-dlp, and node via the system shell. It also executes a bundled Node.js CLI at scripts/opusclip and uses utilities like shasum and brew for environment setup and integrity checks.
[EXTERNAL_DOWNLOADS]: The skill leverages the yt-dlp utility to download video content from well-known social media platforms including YouTube, TikTok, and Instagram for local analysis and editing.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, creating a vulnerability surface for indirect prompt injection. * Ingestion points: External data enters through the opusclip project transcript command and the opusclip_analyze_reference_media MCP tool output. * Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions for the processed transcript text. * Capability inventory: The skill has significant local capabilities including shell execution, file system access, and network communication. * Sanitization: No explicit sanitization or content filtering is described for the external video metadata or transcripts.
[DATA_EXFILTRATION]: The scripts/telemetry.mjs script transmits anonymous usage statistics (such as tool availability, revisions, and aspect ratios) to Mixpanel (api.mixpanel.com). This telemetry is documented and uses a well-known analytics service.

opusclip