tests-prod-run-prod-tests

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to compile reports based on the output of sub-skills that interact with external web content.
  • Ingestion points: The skill processes output from sub-skills tests-prod-test-bitly-links and tests-prod-test-prod-site-links which fetch data from the live site (SKILL.md).
  • Boundary markers: No delimiters or instructions to ignore potential commands within the external data are present in the summary instructions.
  • Capability inventory: The skill uses tools with command execution capabilities, specifically Bash and Task (SKILL.md).
  • Sanitization: The skill lacks logic to sanitize or validate the data ingested from external sources before it is interpreted by the agent and formatted into the final report.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 05:00 PM