skillbox-quickstart

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). This is a raw GitHub shell script (install.sh) served from an unverified user repo and the skill instructs piping it to bash (curl … | bash), which allows arbitrary remote code execution and is a common vector for malware/supply‑chain attacks unless the script and repo are audited and trusted.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and executes third-party code (e.g., the Phase 4 Path A curl to https://raw.githubusercontent.com/build000r/skillbox/main/install.sh) and its generated overlay embeds arbitrary repo "remote" URLs (in scripts/generate_overlay.py -> overlay "source.url") which the workflow will clone/run, meaning untrusted public repository content can influence provisioning and runtime actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill contains a runtime command that fetches and executes remote code via: curl -fsSL https://raw.githubusercontent.com/build000r/skillbox/main/install.sh | bash -s -- --client {CLIENT_ID}, which downloads and runs external script content (remote code execution) as a required install path if the skillbox repo isn't cloned.

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

• Attempt to modify system services in skill instructions detected (high risk: 0.80). The prompt directs the agent to run system-level commands that write config files, install code (curl | bash), copy overlays, and start/ provision boxes/containers which modify the local machine state and execute potentially privileged operations even though it doesn't explicitly request sudo or new users.