Chaos experiment design
Skill shinzoxD/knackbox/skills/coding/chaos-experiment-design
Curated, auditable, benchmark-ready Agent Skills library for Claude Code, Codex, OpenCode, Cursor, and more.
npx -y skills add shinzoxD/knackbox --skill chaos-experiment-designAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 28 days oldThe repository was created 28 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Plan chaos and resilience experiments safely with hypotheses, blast radius limits, and abort criteria. Use whenever the user asks for chaos engineering, game days, fault injection, resilience experiments, or how to safely break things in staging or production.
The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
1.7 KB, 299 tokens by cl100k_base, as published. Nobody here has run it
Chaos Experiment Design
Chaos without a hypothesis is vandalism. Scope blast radius, define steady state, abort criteria, and learnings — prefer staging before prod.
Workflow
- Hypothesis: if X fails, Y still holds (SLO/user journey).
- Steady-state metrics and dashboards.
- Fault: type, target, magnitude, duration.
- Blast radius, blast radius controls, abort conditions.
- Observability and comms (who knows, status page if needed).
- Runbook for inject + stop; post-experiment notes.
- Never surprise prod; approval for prod experiments.
Output format
## Experiment: <name>
**Hypothesis:** …
**Steady state:** …
**Method:** …
**Blast radius:** …
**Abort if:** …
**Schedule / env:** …
**Rollback:** …
**Learning questions:** …
Rules
- No unannounced production chaos.
- One variable at a time when possible.
- Customer-impacting faults need explicit approval language.
- Capture learning even if the system “passed”.
- Do not invent tool-specific syntax you are unsure of.
Edge cases
- Game days: facilitation roles + scenario script.
- Multi-region: partial vs full region loss.
- Data plane faults: disk full, poison messages, clock skew.