The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests and processes untrusted data from multiple sources.
Ingestion points: User inputs provided via materialText, materialUrl, and materialFile parameters as defined in SKILL.md and references/form-schema.json.
Boundary markers: None identified in the provided instructions or scripts; the agent relies on the backend service for instruction isolation.
Capability inventory: The skill is limited to making network POST requests to the vendor's API endpoint (ai-skills.ai/api/execute) via the scripts/run.py script.
Sanitization: No client-side sanitization, filtering, or escaping is performed on user-provided content before transmission.
[SAFE]: Secret management is implemented correctly. The skill requires AISKILLS_API_KEY to be provided via environment variables, avoiding the risk of hardcoded credentials.
[SAFE]: Network activity is restricted to the vendor's infrastructure. The scripts/run.py script communicates exclusively with ai-skills.ai, which is the authoritative domain for the skill provider.

ai-avatar-video