fec-web-video-presentation

Pass

Audited by Gen Agent Trust Hub on Oct 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided content (articles, talk scripts, lesson plans) to generate code for presentation chapters. It lacks explicit instructions for sanitizing this input or using robust boundary markers to prevent malicious instructions embedded within the source text from influencing the agent's code generation logic.
  • Ingestion points: SKILL.md Procedure 1 defines the ingestion of user-submitted articles or long text.
  • Boundary markers: Absent; the skill does not specify delimiters or warnings to ignore instructions inside the processed data.
  • Capability inventory: The skill utilizes scripts/create-video-presentation.mjs to write files to the local filesystem and generates React/TypeScript source code for chapters as described in SKILL.md Procedure 4.
  • Sanitization: Absent; there is no mention of escaping or validating the content extracted from user articles before interpolation into the generated code.
  • [EXTERNAL_DOWNLOADS]: The scripts/create-video-presentation.mjs file generates a package.json containing several external dependencies from the public npm registry. While these are well-known and legitimate packages, they are specified using the latest tag rather than pinned versions, which introduces a dependency on the state of the public registry and could lead to supply chain risks during project initialization.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 28, 2026, 01:10 AM