The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructs the agent to read docs/TOOLS-AND-API-KEYS.md to set up services. This file is explicitly identified as a repository for sensitive configuration and credentials.
[COMMAND_EXECUTION]: The skill performs several command-line operations as part of its primary workflow:
Executes ffprobe to inspect video source streams.
Runs a local Node.js script node scripts/transcribe-elevenlabs.mjs for transcription services.
References various other local scripts and markdown files for sub-tasks like silence trimming and mistake cutting.
[INDIRECT_PROMPT_INJECTION]: The skill processes video transcripts and project briefs which are external, potentially untrusted data sources.
Ingestion points: Raw talking-head video audio (processed into text) and project files in video-projects/<slug>/.
Boundary markers: None identified. There are no instructions for the agent to ignore or delimit instructions found within the transcripts.
Capability inventory: The skill has capabilities for file system access (read/write), local command execution (ffprobe, node), and network service interaction via transcription APIs.
Sanitization: No sanitization or validation of the transcript content is performed before the agent uses it for subsequent editing decisions or beat sheet generation.

edit-video