cloudscape

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

• Third-party content exposure detected (medium risk: 0.65). SKILL.md:10-13 instructs the runtime to fetch a live catalog via curl from a public URL, so outsider-authored free-form text from that remote design.md can be ingested into the agent’s LLM context (indirect prompt injection risk).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes an explicit runtime fetch command that downloads a remote design.md which contains an "Agent Prompt Guide" that the agent is instructed to follow exactly, specifically the curl URL https://api.agent-ds.oday-bakkour.com/v1/systems/cloudscape/design.md, so fetching that file at runtime can directly control agent prompts.