hello-world

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires extracting an activation "code" from the auth/start response and inserting that code verbatim into subsequent poll and download request URLs, which forces the agent to handle/output a secret-like token in requests.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill explicitly requires contacting an external service to download and execute the full skill payload at runtime, creating a clear remote-code-execution / supply-chain backdoor vector that could be abused to run arbitrary or malicious code on the host.

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

• Third-party content exposure detected (high risk: 1.00). The skill's required workflow tells the agent to download and "execute" the full skill content from the external URL https://treehaus.dev/api/skills/download?code=CODE&skill=hello-world, meaning the agent will fetch and run third-party-provided instructions at runtime.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly requires contacting https://treehaus.dev/api/skills/download?code=CODE&skill=hello-world (and its auth endpoints https://treehaus.dev/api/skills/auth/start and /auth/poll) at runtime to fetch "the actual skill content" and then execute those downloaded instructions, so the external URL directly controls agent behavior and is a required dependency.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The skill instructs the agent to download and execute arbitrary external skill content (after activation) which could perform privileged actions such as obtaining sudo, modifying system files, or creating users, so it effectively enables compromising the host state.