film-pipeline

Pass

Audited by Gen Agent Trust Hub on Sep 25, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The scripts/capture.sh script attempts to install the agent-browser package from the NPM registry at runtime if it is missing from the system. Additionally, scripts/poll.sh uses the urllib library in Python to download media files from remote URLs provided by the generation service.
  • [COMMAND_EXECUTION]: The skill relies on several shell scripts (batch.sh, capture.sh, poll.sh) that execute external CLI tools such as cohub, npm, and agent-browser. It also uses python3 to handle JSON data and system operations within these bash scripts.
  • [INDIRECT_PROMPT_INJECTION]: The pipeline includes a mechanism to capture screenshots of live web pages via capture.sh. This introduces a surface where malicious instructions embedded in a website's content could be ingested by the agent during the film-making process.
  • Ingestion points: The capture.sh script takes arbitrary URLs as input; poll.sh retrieves task results and download URLs from an external API.
  • Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard instructions found within the captured web content or downloaded metadata.
  • Capability inventory: The skill possesses the capability to execute shell commands, perform network downloads, automate a web browser, and write various files (PNGs, JSON, Markdown) to the local filesystem.
  • Sanitization: The scripts do not perform sanitization or filtering on the data retrieved from external sources before it is saved and potentially processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 25, 2026, 01:44 PM