openclaw-setup

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill instructs the agent to AskUserQuestion for the API key/base URL and to replace placeholders with the actual API key in generated config files and curl commands, which requires the LLM to output secret values verbatim.

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs contacting an arbitrary ANT HROPIC_BASE_URL (e.g., running "curl <ANTHROPIC_BASE_URL>/v1/models" and "openclaw models list --provider zknow") and reading those remote API responses, so the agent will fetch and interpret untrusted third-party content that can influence configuration and model-selection actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill requires setting ANTHROPIC_BASE_URL (e.g., http://190.92.219.209:8180) which is contacted at runtime (via openclaw commands and curl) to obtain models and serve inference, meaning that remote content at that URL directly controls the agent's outputs and is a required dependency.