broll-director

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill is instructed to run a local JavaScript file validate-shot-recipes.mjs to verify generated JSON artifacts during the delivery phase.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data including SRT files, user goals, and audience profiles, creating an attack surface for embedded malicious instructions.
  • Ingestion points: SRT files, mode, talking-head video, user goal, audience, platform, and output constraints defined in SKILL.md.
  • Boundary markers: The instructions do not specify any delimiters or safety markers to isolate the untrusted SRT content from the agent's core logic.
  • Capability inventory: The skill has the capability to write JSON/Markdown files to the local file system and execute a local validation script.
  • Sanitization: The skill contains a minor instruction to use 'safe generic wording for unconfirmed transcript facts' but lacks technical sanitization or filtering for the external inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 07:43 AM
Security Audit — agent-trust-hub — broll-director