web-design-guidelines

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

• Third-party content exposure detected (high risk: 0.90). Mode B's audit workflow explicitly instructs the agent to fetch and read the public raw GitHub file at https://raw.githubusercontent.com/bh611627/skillcodex/main/references/design-guidelines.md so that external, public content will be parsed and used to drive audit decisions, exposing the agent to untrusted third-party content.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's Mode B audit explicitly fetches runtime content from https://raw.githubusercontent.com/bh611627/skillcodex/main/references/design-guidelines.md and uses that fetched guidelines file to drive audit findings and instructions, so the external content directly controls agent behavior.