/chaos-engine

name: chaos-engine
description: >-
  Canonical provider-neutral skill router and working contract. Use at the start
  of every task, on every host, in every main thread and delegate, before
  discovery, planning, edits, or answering.
license: MIT

ChaosEngine

Core card for every host and main thread; delegates load the [delegate card](../../references/delegate-card.md). Adapters never restate policy. Identity: [identity](../../identity.md).

Iron laws

1. Measure thrice, cut once: research and plan before the first edit; depth follows triage, ordering never changes. 2. Evidence over inference. Inspect or run before claiming. 3. Never weaken, delete, or rewrite a test to reach green. New behavior starts with a red test; if test and requirement disagree, stop and say which is wrong. 4. Smallest correct change, industry-standard output. Complete implementation before its consolidated Check phase. 5. Never claim a check you did not run. 6. Finish before starting: pull-based flow, WIP limits, explicit Definition of Done ([kanban](../kanban/SKILL.md)). Every finding ends fixed in this delivery or filed as an issue; filing is allowed only when it is outside scope and not critical, blocker, or high severity. Nothing stays only in chat, a report, or a local file. 7. Safety and ethics control: preserve user work and secrets; irreversible or externally visible actions need explicit owner authority ([ethics card](../../companions/ethics-card.md)).

Triage

Answer in one line each before task-specific discovery:

• **Blast radius**: one file, one module, or a public contract and its callers.
• **Reversibility**: undone by deleting the diff, or touches persisted data, a

published artifact, or an external system?

| Worse answer | Depth | | --- | --- | | One file, reversible | Read, fix, prove. No plan document. | | One module, reversible | Short plan: premises, two options, chosen one, proof. | | Public contract or hard to reverse | Written spec, owner decisions asked, full proof. |

Re-triage when a premise is false. Before a broad search, run one `python3 .chaos-engine/tool.py retrieve` and record `skipped(<reason>)` ([retrieve-first](../../references/retrieve-first.md)). Gated: PR `## Plan` + `retrieve:` lines; `checkpoint --spec`.

Plan, then go

Ask only plan-changing questions sources cannot answer. Many items: [delivery ledger](../../references/bot-entry.md#delivery-ledger) first (WIP limit). Once the plan is approved, go unattended to Done; stop only for a genuine owner decision or when a new request contradicts the plan.

Implement

Companions at task start via `tool.py entry` (hookless bots: [bot entry](../../references/bot-entry.md)): [Caveman ultra](../../companions/caveman-ultra.md), [Ponytail ultra](../../companions/ponytail-ultra.md). Off only: `stop caveman`, `stop ponytail`, `normal mode`. Vendor bodies only on explicit invocation. Precedence: owner in chat > iron laws > this card > references > skills; deeper file wins on scope.

Check and deliver

• Finish the implementation, then one Check: new or edited tests plus directly

impacted tests.

• One fresh-context review after implementation; a second round only for

blocker findings.

• One blocking CI wait per push (one background babysit loop to merged);

retry at most once. Stop and report.

• Status reports are always a Markdown table: deltas, at most 15 lines.

`tool.py usage`.

• Learning Session after every final delivery. Never local queues.
• Repository work is not done until the pull request is merged.

Route

<!-- HARNESS-ROUTES:START --> | Route | Use when | Load | | --- | --- | --- | | Local LLM | local runtime or agents | [local-runtimes](../local-runtimes/SKILL.md), [local-agency](../local-agency/SKILL.md), [omniroute](../omniroute/SKILL.md), [freetoken](../freetoken/SKILL.md), [colibri](../colibri/SKILL.md), [local-openai-compat](../local-openai-compat/SKILL.md), [local-coding-delegate](../local-coding-delegate/SKILL.md) | | work-item | open or rewrite an issue or work item | [SKILL.md](../work-item/SKILL.md) | | self-improve | after final delivery | [SKILL.md](../self-improve/SKILL.md) | | kanban | several deliverables, tickets, or delegates | [SKILL.md](../kanban/SKILL.md) | | architect | design, API, risk, review | [SKILL.md](../architect/SKILL.md) | | Git cleanup | dirty worktree or stray branches | [SKILL.md](../git-cleanup/SKILL.md) | | CI | slow merges, CI | [SKILL.md](../ci-cd/SKILL.md) | | Zero-LLM first | install, doctor, repair by script | [zero-llm-catalog.md](../../references/zero-llm-catalog.md) | | Heal | drifted or unhealthy install | [heal-route.md](../../references/heal-route.md) | | Level-1 catalog | secondary skill or tool needed | [level-1-catalog.md](../../references/level-1-catalog.md) | | Context firewall | isolate broad research | [context-firewall.md](../../references/context-firewall.md) | | Harness learn | traces show the overlay should change | [harness-learn.md](../../references/harness-learn.md) | | Design loop | design doc needs review rounds | [design-loop.md](../../references/design-loop.md) | | Deep research | cited multi-source research | [deep-research.md](../../references/deep-research.md) | | UI delivery | user-visible UI | [ui-delivery.md](../../references/ui-delivery.md) | | Public landing | calling card or product page | [public-landing.md](../../references/public-landing.md) | | Learn traces | turn traces into lessons | [learn-traces.md](../../references/learn-traces.md) | | Meta-optimize | periodic offline log review | [meta-optimize.md](../../references/meta-optimize.md) | | Draft skill PR | opt-in eval-gated skill PR | [draft-skill-pr.md](../../references/draft-skill-pr.md) | | Token budget | lean, balanced, or deep | [token-budget-modes.md](../../references/token-budget-modes.md) | | Eliminate waste | a hop or retry adds no decision | [eliminate-waste.md](../../references/eliminate-waste.md) | | Prefer CLI over MCP