Chaos Engineer

What it does

When your agent uses it

• Designing a chaos experiment with a hypothesis, steady state and blast radius
• Implementing failure injection with Chaos Monkey or Litmus
• Planning and running a game day exercise
• Adding continuous chaos testing to CI/CD
• Testing Kubernetes pod and node failures

Example prompts

• “Design a chaos experiment for our checkout service that kills one pod and watches latency.”
• “Plan a game day for a region failover, including rollback steps and a post-mortem template.”
• “Add a chaos test stage to our CI/CD pipeline with a safe blast radius.”

Workflow steps

4 steps, taken from the step headings in SKILL.md.

1. 1Define steady state and apply the experiment
2. 2Create and apply a Litmus ChaosEngine manifest
3. 3Monitor during the experiment
4. 4Rollback / abort if steady state is violated

What it can do on your machine

Read from SKILL.md and the folder at commit 1be15d8. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • kubectl
  • brew

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  Links to these hosts (documentation or services it may open):

  • github.com
  • synergetic.solutions
  • jeffallan.github.io

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Chaos Engineer loads about 1.8k tokens when it runs, and up to ~15k if it reads all its reference files. Until then it costs about 124 tokens; SKILL.md has 368 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~124

When it runs · the whole SKILL.md, loaded when a task matches

~1.8k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~15k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from Jeffallan/claude-skills at commit 1be15d8, republished under its MIT licence (© Jeffallan). 368 words, ~1,759 tokens.

name

chaos-engineer

description

Designs chaos experiments, creates failure injection frameworks, and facilitates game day exercises for distributed systems — producing runbooks, experiment manifests, rollback procedures, and post-mortem templates. Use when designing chaos experiments, implementing failure injection frameworks, or conducting game day exercises. Invoke for chaos experiments, resilience testing, blast radius control, game days, antifragile systems, fault injection, Chaos Monkey, Litmus Chaos.

license

MIT

metadata.author

https://github.com/Jeffallan

metadata.company

https://synergetic.solutions

metadata.version

1.1.0

metadata.domain

devops

metadata.triggers

chaos engineering, resilience testing, failure injection, game day, blast radius, chaos experiment, fault injection, Chaos Monkey, Litmus Chaos, antifragile

metadata.role

specialist

metadata.scope

implementation

metadata.output-format

code

metadata.related-skills

sre-engineer, devops-engineer, kubernetes-specialist

# Verify baseline: p99 latency < 200ms, error rate < 0.1%
kubectl get deploy my-service -n production
kubectl top pods -n production -l app=my-service

# chaos-pod-delete.yaml
apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: my-service-pod-delete
  namespace: production
spec:
  appinfo:
    appns: production
    applabel: "app=my-service"
    appkind: deployment
  # Limit blast radius: only 1 replica at a time
  engineState: active
  chaosServiceAccount: litmus-admin
  experiments:
    - name: pod-delete
      spec:
        components:
          env:
            - name: TOTAL_CHAOS_DURATION
              value: "60"          # seconds
            - name: CHAOS_INTERVAL
              value: "20"          # delete one pod every 20s
            - name: FORCE
              value: "false"
            - name: PODS_AFFECTED_PERC
              value: "33"          # max 33% of replicas affected

# Apply the experiment
kubectl apply -f chaos-pod-delete.yaml

# Watch experiment status
kubectl describe chaosengine my-service-pod-delete -n production
kubectl get chaosresult my-service-pod-delete-pod-delete -n production -w

# Tail application logs for errors
kubectl logs -l app=my-service -n production --since=2m -f

# Check ChaosResult verdict when complete
kubectl get chaosresult my-service-pod-delete-pod-delete \
  -n production -o jsonpath='{.status.experimentStatus.verdict}'

# Immediately stop the experiment
kubectl patch chaosengine my-service-pod-delete \
  -n production --type merge -p '{"spec":{"engineState":"stop"}}'

# Confirm all pods are healthy
kubectl rollout status deployment/my-service -n production

# Install toxiproxy CLI
brew install toxiproxy   # macOS; use the binary release on Linux

# Start toxiproxy server (runs alongside your service)
toxiproxy-server &

# Create a proxy for your downstream dependency
toxiproxy-cli create -l 0.0.0.0:22222 -u downstream-db:5432 db-proxy

# Inject 300ms latency with 10% jitter — blast radius: this proxy only
toxiproxy-cli toxic add db-proxy -t latency -a latency=300 -a jitter=30

# Run your load test / observe metrics here ...

# Remove the toxic to restore normal behaviour
toxiproxy-cli toxic remove db-proxy -n latency_downstream

# chaos-monkey-config.yml — restrict to a single ASG
deployment:
  enabled: true
  regionIndependence: false
chaos:
  enabled: true
  meanTimeBetweenKillsInWorkDays: 2
  minTimeBetweenKillsInWorkDays: 1
  grouping: APP           # kill one instance per app, not per cluster
  exceptions:
    - account: production
      region: us-east-1
      detail: "*-canary"  # never kill canary instances

# Apply and trigger a manual kill for testing
chaos-monkey --app my-service --account staging --dry-run false

© Jeffallan, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 5 other files (references) in skills/chaos-engineer of Jeffallan/claude-skills.

• SKILL.md
• references/chaos-tools.md
• references/experiment-design.md
• references/game-days.md
• references/infrastructure-chaos.md
• references/kubernetes-chaos.md

Open the folder on GitHubat commit 1be15d8