Create verification strategy

Skill qa-veritas/skillpack/skills/create_verification_strategy

Agent capability that loads only when the task needs it. The Skills layer of QA Veritas.

Install

npx -y skills add qa-veritas/skillpack --skill create_verification_strategy

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

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

Given an intended change, define the pre-check, the observable post-checks that prove it worked, and the signal that triggers rollback. Use before applying any change to a live resource.

SKILL.md

2.7 KB, as published. Nobody here has run it

create_verification_strategy

Purpose

A change is not done when it is applied; it is done when an observable signal confirms it. This skill turns an intent ("raise the index cluster heap to 16 GB") into the checks that prove success and the signal that says "roll back now," before the change is made.

Inputs

intended_change — the action and its goal, in plain terms.
current_state (optional) — the relevant facts from inventory (current values, capacity, dependents).

Outputs

pre_check:
  - "Heap currently 8 GB; node RAM 32 GB; rule: heap <= 50% RAM, <= 31 GB."
  - "16 GB satisfies the rule; proceed."
post_checks:
  - check: "Process reports -Xmx16g"
    signal: "JVM args show 16g"
  - check: "Cluster returns to green within 5 min"
    signal: "health endpoint status == green"
  - check: "No OOM / restart loop"
    signal: "0 restarts over 10 min"
rollback_signal: >-
  Cluster not green within 5 min, OR any restart, OR heap pressure
  alarms. Rollback path: re-create with previous -Xmx and keep the old
  container as a named fallback until green.

Prompt template

You are designing the verification for a change before it is made. Inputs: intent {intended_change}, current state {current_state}.

Write a pre-check that confirms the change is feasible and safe against recorded capacity and rules of thumb. If it fails, stop and recommend the smaller change.

Define post-checks that are observable — a health status, a port listening, a process arg, a reconverged count. Not "looks fine." Each post-check names the exact signal.

Define the rollback signal and the rollback path. Prefer reversible actions (keep the previous artifact as a named fallback until the post-checks pass). Output the structured strategy. The change itself is out of scope — this skill only defines how we will know it worked.

Examples

Resize a data-bearing VM. Pre-check: capacity + graceful path (ACPI shutdown, never hard kill). Post-check: VM boots, service rejoins, data intact. Rollback: revert size, power on.
Add a proxied hostname. Post-check: config test passes and the new host resolves to the backend. Rollback: remove the server block, reload.

Composition

Consumes identify_root_cause (so the fix verification targets the real cause) and uses analyze_cluster_health / check_disk_pressure as post-checks.