Quality gate for AI-generated code

AI writes the code.
You set the standard.

Normik checks every pull request against your team's playbook — your rules, your API key, your CI. No app to install. No code stored.

Playbook check

CI

error src/api.ts:42
Direct fetch — use the internal HTTP client.

1 blocking violation — merge gated

Claude Code

MCP

Loaded 8 rules · fixed 2 violations
before committing

Pre-commit check passed
Pending reviewAI proposed

Migrations always ship a rollback

ApproveReject

Runs on your key

  • Anthropic
  • OpenAI
  • Mistral
  • Google

Fits your workflow

  • GitHub Actions
  • GitLab CI
  • Claude Code
  • Any MCP agent

Generic AI review tells you what it thinks.Normik enforces what you decided.

Your playbook, not generic rules

Write your conventions in plain language — “HTTP calls go through our internal client” — versioned, scoped by glob, tagged, and governed. Or bootstrap it from your repo in one call.

A CI gate with nothing to install

One job: git diff piped to Normik. Blocking violations fail the check and get posted as a PR comment by your own CI token. No app, no webhook, and if Normik is down, your merges aren't.

Agents fix issues before the commit

Claude Code and any MCP agent load your rules while they write, self-check the diff, and report what they fixed. Quality shifts left of the pull request.

AI proposals, human approval

Agents can propose new rules and skills from what they see in your codebase. Everything AI-generated is flagged and waits in a review queue until a human approves it.

Compliance you can show a client

Compliance rate, most-violated rules, pre-commit fix rate, token spend, audit log. Per client workspace — built for agencies managing many codebases.

Your key. Your provider. Your code.

Analysis runs on your own API key — Anthropic, OpenAI, Mistral or Google. Keys are encrypted, code is never stored, and the token bill stays yours.

Live in an afternoon.
No app. No webhook. No code leaves your CI.

01

Write — or bootstrap — your playbook

Add your team's rules in plain language, or let your agent submit a sample of the repo and derive a first playbook for review.

02

Add one job to your CI

Paste the snippet into GitHub Actions or GitLab CI. Every pull request is checked against your rules, on your key.

03

Watch drift become visible

Compliance per workspace, violations by severity, what agents fixed before committing. Governance, not vibes.

# .github/workflows/normik.yml — the whole integration
- run: |
    git diff "origin/main...HEAD" |
    curl -sS --fail-with-body -X POST "$NORMIK_URL/api/ci/gate" \
      -H "Authorization: Bearer $NORMIK_TOKEN" --data-binary @-

Starting from a blank playbook is the hard part — so we wrote seven of them. Browse the playbook library →

Unlimited analyses on every plan.

Analysis runs on your own API key, so we never meter it. Pay for organization, not for usage.

Free

€0

Unlimited analyses on your key · 1 workspace · CI gate + MCP agents

Get started

Team

€39/month

3 workspaces · 10 members · rule & skill versioning · 14-day trial

Start free trial

Agency

€149/month

Unlimited client workspaces · up to 50 members · AI governance · audit log · analytics

Start free trial

Need more seats? Enterprise — €349/month →

Your agents already code fast. Make them code like your team.

Start free — no card required