use-railway

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt instructs the agent to fetch and relay credentials and one-time sign-in codes verbatim (e.g., bucket access/secret keys via railway bucket credentials, device-code links/codes, and railway variable set KEY=value usage), which requires the LLM to output secret values directly.

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

• Third-party content exposure detected (high risk: 0.75). The skill runs scripts/analyze-mongo.py, which calls get_recent_logs(service, ...) to fetch runtime log lines from the Railway service and then includes them in the LLM context as recent_logs/recent_errors (outsider-authored free text: application/runtime logs produced by other parties).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes explicit install commands that fetch and execute remote shell code at runtime (bash <(curl -fsSL https://railway.com/install.sh)), which would run code from https://railway.com/install.sh as a required installation step for the CLI.