Skip to content
Development
Skill

/verification-harness

Create or audit a deterministic project-local verification harness inside a caller-owned Verify phase. Direct use is audit-only. Do not use only to run an existing check.

BOOST
From plugin
khanelinix
34750 skills10 agents6 commands
Install
$ npx -y skills add khaneliman/khanelinix --skill verification-harness --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/verification-harness

Context preview

The summary Claude sees to decide when to auto-load this skill.

Create or audit a deterministic project-local verification harness inside a caller-owned Verify phase. Direct use is audit-only. Do not use only to run an existing check.

SKILL.md

verification-harness.SKILL.md
name: verification-harness
description: "Create or audit a deterministic project-local verification harness inside a caller-owned Verify phase. Direct use is audit-only. Do not use only to run an existing check."
license: Complete terms in LICENSE

Verification Harness

Create or audit one project-local verification surface. Keep the surface deterministic, portable, observable, minimal, and rerunnable.

Modes

  • **Create.** Build or repair a harness only when the caller explicitly requests

or authorizes the required writes, including necessary verification work within an already-authorized implementation. Do not request permission again for that scope or add broad infrastructure when a focused check suffices.

  • **Audit.** Inspect an existing harness, exercise its real outcomes, and list

maintenance and coverage gaps without changing it. If the harness is missing, return a proposed feature/check map and exact write scope.

Workflow

1. Confirm project scope, Create or Audit mode, supported environments, and exit criteria. Stop when the request mixes modes without a clear boundary. 2. Build a feature/check map. Record each behavior, observable outcome, minimal command, fixture or input, owner, and known gap. 3. In Create mode, add the smallest portable commands and fixtures that prove real outcomes. Prefer project-native tools and deterministic test data. 4. In Audit mode, run each mapped check and compare expected outcomes with observed outcomes. Mark missing, flaky, environment-bound, or stale checks. 5. Re-run changed checks from a clean invocation. Record commands, exit status, observable evidence, and unverified conditions. 6. After behavior changes, update the feature/check map and affected checks. Do not claim coverage when the map or maintenance evidence is stale.

Output

Return the selected mode, feature/check map, changed or audited commands, observable results, rerun commands, maintenance actions, and explicit gaps. Separate harness creation or repair from ordinary check execution.

If the caller only asks to run an existing check, return that request to the caller's normal verification method. This skill does not own ordinary check execution.

Boundaries

The caller owns lifecycle, architecture, final judgment, and external writes. Keep this skill focused on the verification surface and its evidence.

Read more
Ships withkhanelinix

Run packages directly with: Or install from the packages output. For example:

Get the whole plugin

Other skills on khanelinix.