autocli

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). The prompt instructs executing a remote shell script via curl from an unvetted GitHub user (raw.githubusercontent.com/nashsu/AutoCLI/main/scripts/install.sh), which is a high‑risk pattern for malware distribution (the GitHub repo owner appears unverified and distributing binaries/scripts directly is unsafe); the GitHub repo link likewise could host malicious installers while the X status URL is benign but irrelevant.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to fetch and scrape live, user-generated content from many public third-party sites (e.g., SKILL.md and README.md show commands like "autocli reddit read --url <post_url>", "autocli twitter timeline", "autocli youtube search", and the YAML/browser_evaluate adapter workflow for arbitrary URLs), so untrusted web content will be read and can influence the agent's actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill instructs at runtime to install a required dependency by running remote shell code via curl -fsSL https://raw.githubusercontent.com/nashsu/AutoCLI/main/scripts/install.sh | sh, which fetches and executes remote code that the skill depends on.