The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation in SKILL.md provides a setup command to install the HeyGen CLI by downloading a script from https://static.heygen.ai/cli/install.sh and piping it to bash. This is a standard installation method provided by the official vendor for the required dependency.
[COMMAND_EXECUTION]: The skill interacts with system binaries (ffprobe and the heygen CLI) using execFileSync with explicit argument arrays. This prevents shell injection vulnerabilities by ensuring that user-provided search intents and file paths are handled as literal arguments rather than part of a shell command string.
[INDIRECT_PROMPT_INJECTION]: The skill generates an inventory ledger (index.md) that incorporates potentially untrusted external data.
Ingestion points: User search intents and metadata returned from the HeyGen API.
Boundary markers: Absent in the generated Markdown index table.
Capability inventory: File system write operations, subprocess execution via system binaries, and network downloads via fetch.
Sanitization: Input data is reflected in the manifest and index files without additional escaping or sanitization, though the impact is limited by the non-executable context of the ledger.
[DATA_EXFILTRATION]: The skill manages a content-addressed global cache in the user's home directory (~/.media/) and communicates search queries to the provider API. This behavior is documented and appropriate for the skill's primary function of media management and reuse.

media-use