ethanplusai/codex-on-crack / plugins/codex-on-crack/skills/crack
crack
UnverifiedLead substantial Codex builds with one configured implementation worker and lead acceptance. Use when the user requests codex-on-crack, names a model to delegate work to, asks to run subagents through Claude, or requests a lead/worker split. Keep host-only work direct; honor explicit external model choices. Workers must not invoke this orchestrator.
references/verification.md
View on GitHub# Choosing verification
Read this when you are deciding what evidence a change needs. The goal is an
honest check of the actual behavior, not a fixed count of stages.
## Match evidence to the feature
- Identify the observable result the user asked for and check that journey, not
only an internal function.
- Cover the negative and boundary cases that could silently change behavior:
invalid input, empty or missing data, ordering, limits, and error paths.
- Check integration points and any real UI state the change touches. A screenshot
or a driven browser session is evidence; a claim that it "should render" is not.
- When the feature persists data, exercise the failure boundaries: a rejected
write, a partial failure, a concurrent change, and recovery after restart.
Load this only for persistence work, not as boilerplate for every task.
## Prefer the smallest honest check
Run the targeted test for the behavior you changed. Widen to the full suite when
shared code, packaging, configuration, or a release inventory makes the wider run
the honest check. Do not rerun an entire suite after every small edit, and do not
call a passing test suite proof of a user journey it never exercised.
## Keep evidence separable from claims
Record each command with its working directory, exit status, and the result that
matters. When a check cannot run, say so and why. Missing evidence is missing,
never a pass, and a worker report is an index to the artifacts, not a substitute
for inspecting them.
## Blocking findings
A confirmed defect that can lose or corrupt data, or that breaks authorization or
a security boundary, blocks acceptance even when every automated check passes.
Report it with its evidence instead of trading it against a passing suite, and
keep looking for the same class of problem when the change touches persistence,
permissions, or tenancy.
## Independent scrutiny
Use a second independent pass when a concrete risk or inadequate evidence
justifies it, not by default. Fresh context on the same model can help; a
different model is not automatically a better reviewer.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |