dd-apm

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and executes third-party content (e.g., linux-ssi/agent-install Phase 2 runs curl -L https://install.datadoghq.com/scripts/install_script_agent7.sh and executes it) and repeatedly ingests and interprets Datadog backend output via the pup CLI (e.g., pup apm troubleshooting list, pup traces search) as part of its decision and remediation workflow, so untrusted external content can materially influence tool use and next actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill fetches and executes remote installer code at runtime (notably the Datadog install script: https://install.datadoghq.com/scripts/install_script_agent7.sh is curl'ed and run via bash, and it also relies on remote Helm charts via https://helm.datadoghq.com and pup GitHub release downloads), so these external URLs are required by the skill and will execute remote code on the target.

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 Datadog operator, creating secrets and Kubernetes resources, applying manifests and restarting pods (cluster-level changes that modify system state and require elevated/kubeadmin privileges), so it directs actions that can change or compromise the machine/cluster state.