rocket-launch-to-reel
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external data from mission reports and web searches to generate scripts and captions. This creates a surface where malicious instructions could be embedded in the source material (e.g., a web page describing a launch).
- Ingestion points: Uses
WebSearchandWebFetchto retrieve information about vehicles, payloads, and launch outcomes. - Boundary markers: The workflow specifies a "Verify the launch" step requiring cited sources, which acts as a manual/logical check against misinformation.
- Capability inventory: The skill only produces a structured JSON output (hook, script, captions). It does not have access to file-system writes, shell execution, or outbound POST requests beyond the defined tool set.
- Sanitization: The output is restricted to a specific JSON contract, limiting the impact of potential injections to the content of the script itself rather than system-level compromise.
Audit Metadata