The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions include a setup step to install the omni-captions-skills Python package using a custom index URL hosted on the author's GitHub Pages site (https://lattifai.github.io/pypi/simple/).
[COMMAND_EXECUTION]: The skill uses the Bash(omnicaptions:*) tool to perform operations like omnicaptions transcribe and omnicaptions convert via the command line.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, which could contain malicious instructions intended to influence the agent's behavior.
Ingestion points: Untrusted data enters the agent's context through remote video URLs provided to the transcription command and via various subtitle files (SRT, VTT, ASS, JSON) processed by the conversion command as shown in SKILL.md.
Boundary markers: There are no instructions or delimiters specified in SKILL.md to indicate that embedded content within caption files should be ignored or treated as data only.
Capability inventory: The skill possesses the capability to execute shell commands via Bash(omnicaptions:*) and perform file system operations (read/write) across all conversion tasks.
Sanitization: There is no mention of sanitization, validation, or escaping of the content ingested from external caption formats or transcription sources.

omnicaptions-convert