volcengine-deploy

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

• Third-party content exposure detected (medium risk: 0.30). Runtime path reads user-supplied repo content by resolving a provided Git URL (clone) or local directory and then generating a deployment plan/packaging based on files under that repo (e.g., .env.example, .volcengine/*, source code), so the agent ingests outsider-authored free text from arbitrary repositories.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill clones arbitrary Git URLs at runtime (see the command "git clone --depth 1 "$input" "$work_dir/src""), which fetches remote code that may be built/executed as part of the deployment flow.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill instructs writing files under /opt and managing/starting systemd-backed services (e.g., "binary + systemd" and writing /opt//.env), which are system-level changes that typically require elevated privileges and therefore risk modifying the host machine state.