ai-qa-weekly
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources such as technical forums and GitHub repositories. While this creates a potential for indirect prompt injection, the skill effectively mitigates this via mandatory verification gates that require the agent to independently confirm dates and sources against primary documentation before including any item. This behavior is consistent with the skill's primary function of news curation.\n- [COMMAND_EXECUTION]: The README includes a standard instruction for users to manually symlink the skill directory for local discovery. This is a common, non-malicious configuration step for developers.\n- [SAFE]: The skill instructions emphasize factual integrity and forbid the use of unverified claims or model memory, ensuring the generated reports are source-backed and accurate.
Audit Metadata