The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous scripts and commands intended for execution on infrastructure, including resource exhaustion, network latency injection, and container termination.
Evidence: references/infrastructure-chaos.md contains commands for stress-ng, iperf3, and pumba to saturate CPU, memory, and network resources.
Evidence: references/chaos-tools.md includes a bash script that uses aws ec2 terminate-instances to simulate Chaos Monkey behavior.
[PRIVILEGE_ESCALATION]: Several provided scripts require administrative privileges (sudo) to modify system configurations or install tools.
Evidence: references/infrastructure-chaos.md uses sudo apt-get install -y stress-ng to install testing tools.
Evidence: references/infrastructure-chaos.md uses sudo sh -c 'echo "127.0.0.1 {domain}" >> /etc/hosts' to simulate DNS failures by modifying system network files.
[EXTERNAL_DOWNLOADS]: The skill automates the download and application of Kubernetes manifests from remote sources.
Evidence: references/chaos-tools.md contains a command to kubectl apply a manifest directly from https://litmuschaos.github.io/litmus/litmus-operator-v2.14.0.yaml.
[INDIRECT_PROMPT_INJECTION]: The skill features an attack surface for indirect injection by processing data from external monitoring systems and infrastructure APIs without explicit sanitization patterns.
Ingestion points: references/chaos-tools.md (Prometheus API), references/kubernetes-chaos.md (Kubernetes API).
Boundary markers: Absent.
Capability inventory: references/infrastructure-chaos.md (subprocess.run), references/chaos-tools.md (sh execution in Jenkins/GitHub Actions).
Sanitization: Absent.

chaos-engineer