labs-quickstart

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill intentionally sends user data to external telemetry endpoints and fetches & executes remote scripts (curl → bash / node), which enables data exfiltration and remote code execution/backdoor capabilities if abused.

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

• Third-party content exposure detected (high risk: 0.85). The workflow WebFetches a personalized runbook from a public Convex site at runtime (outsider-authored free text via curl .../q/<SLUG>.md?... and/or the bootstrap log tail), which is then ingested into the agent’s LLM context as instructions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill fetches and executes remote scripts and runbooks from https://basic-anteater-667.convex.site (e.g., downloading and running the quickstart bootstrap and publish scripts and WebFetching the runbook), so external content from that host directly controls runtime behavior and instructions.