The fastest path to understanding Nerviq is not to learn every concept up front. It is to detect the platforms, score the repo, expose the biggest gaps, fix the basics, check drift, and compare the improvement.
Use the global install if you plan to keep Nerviq around. For one-off experiments, `npx @nerviq/cli` works too.
Requires Node.js 18+. The six steps below assume you are already inside the repo you want to inspect.
Need the shortest terminal-only command map first? Run nerviq --beginner to show only audit, setup, fix, augment, and doctor.
Start by seeing which agent surfaces Nerviq can actually find in the repo.
This tells you whether you are dealing with one platform or a drift-prone multi-platform setup before you change anything.
You should see the detected platforms and a Harmony average. Even a single-platform repo is fine here.
Capture the baseline score and save the first snapshot so improvement can be measured later.
Without the first snapshot, `compare` has nothing meaningful to show you later.
You should get a live audit score and a saved audit snapshot for the current repo state.
Expand from the lightweight score to the full recommendation surface.
The short score tells you where you stand. The full audit tells you what to fix first.
Look at critical checks, top next actions, and weakest categories. Do not try to fix everything at once.
Bootstrap the starter-safe layer, then remove the most obvious critical issues.
This is the fastest way to turn a weak baseline into a governed baseline without hand-authoring every file.
You should see generated config files, hooks, rules, commands, and any auto-fixable criticals closed out.
Re-run Harmony now that the repo has a stronger config layer.
This shows whether the repo is converging toward one coherent multi-agent posture or still drifting.
Use this step to verify that platform surfaces feel more aligned than they did in step 1.
Save the second snapshot, then compare it against the first one.
This closes the loop. Nerviq should not just tell you what is wrong. It should show that the repo actually improved.
You should now see the latest score, the previous score, and a concrete delta between the two snapshots.
At this point you should have a baseline, at least two snapshots, a clearer view of drift, and a repo that is materially stronger than when you started. Only then is it worth going deeper into the rest of the product surface.
Use `nerviq plan` and `nerviq apply --dry-run` when you want reviewable rollout instead of direct setup and fix flows.
Go deeper with Harmony docs after the six steps. Harmony is the main surface for cross-platform drift detection and sync decisions.
Inspect the published before/after proof repo to see a full sample baseline, generated after state, and raw audit artifacts.