security-advisory-review

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.70). The workflow instructs the agent to clone a remote repository at runtime and then write/read source code from that repository into its working context for analysis; that repository content is outsider-authored if the repo owner isn’t the operating user’s organization.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs cloning a user-supplied remote repository at runtime (git clone /tmp/sec-review-), which means arbitrary fetched code from the provided URL () can be executed as part of the PoC workflow.