gremlin-chaos

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill instructs users to pass secrets directly on the command line / via helm --set (e.g., --api-token , --set ...teamSecret=), which requires secret values to be embedded verbatim and risks exfiltration or exposure; (the curl example uses an env var which is fine).

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

• Potentially malicious external URL detected (high risk: 0.90). I flagged runtime dependencies that install or trigger remote-executing Gremlin components: the Helm repo URL https://helm.gremlin.com (used by "helm repo add" / "helm install") fetches and deploys remote charts that execute code, and the API endpoints https://api.gremlin.com/v1/attacks/new and https://api.gremlin.com/v1/scenarios/${{ vars.SCENARIO_ID }}/runs are invoked at runtime to trigger attacks (remote execution/control of agents).

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

• Attempt to modify system services in skill instructions detected (high risk: 0.90). The skill instructs installing a host agent using sudo and describes/automates host-level destructive attacks (e.g., shutdown, process killer) and helm/secret creation, which modifies system state and requires elevated privileges.