Stop rebuilding your agent setup on every repo.
You've written the same code-review sub-agent three times. You've copy-pasted a hook that never fires. Agent Ops Kit is the config pack you'd have built yourself over a long weekend — sub-agents, hooks, slash commands, and CLAUDE.md starters you drop into a repo and use the same hour.
Works with Claude Code and other agentic coding tools.
The part nobody warns you about
Nobody sells the boring layer.
- Every new repo starts from zero. You open last month's project, lift half a config out of it, then rewrite most of it anyway because that repo's conventions aren't this repo's conventions.
- Your hooks are decoration. They either never fire or they fire on everything, so you commented them out and forgot. A pre-commit secret gate you don't trust is worse than none.
- You can't tell if it's working. No idea whether your setup is saving context or quietly burning it, and no baseline to compare against. So you keep tinkering instead of shipping.
You're expected to invent it, alone, per repo, forever.
What's inside
| Asset | Count | What it does |
|---|---|---|
| Sub-agent definitions | 8 | code-review, test-backfill, dependency-upgrade, incident-triage, database-migration-writer, flaky-test-hunter, api-contract-checker, changelog-writer — each with a defined job and a stated failure mode |
| Slash commands / skills | 10 | Argument handling and a worked example in every one |
| Hooks | 6 | Secret-scan pre-commit gate, test-on-edit, notify-on-stop, permission allowlist, and two more |
CLAUDE.md starter templates |
5 | TS/Next, Python/FastAPI, Go, Rails, monorepo |
| Agent Ops Playbook | 1 | Context budgeting, orchestration patterns, token-cost control, and the anti-patterns that cost you a re-run |
| Cost-control cheatsheet | 1 | One page. Print it, tape it up. |
Eight sub-agents, not forty. Each one has a job it does and a job it refuses.
Markdown, shell scripts, and JSON you can read before you run. No installer, no account, no server. Unzip into your repo and it works.
Every hook executed with real event JSON (6/6); all 8 sub-agents discovered by the real CLI; all 3 free-repo agents behavior-verified on a real repo; secret-scan proven with positive and negative controls.
Pricing
One purchase. No subscription.
Launch code LAUNCH25 takes 25% off. It expires 48 hours after launch and does not come back.
-
Solo
$19 $14.25
- Included: Full kit
- Included: 1 seat
- Not included: Team conventions doc
- Not included: Reuse in client work
-
Team
$49 $36.75
- Included: Full kit
- Included: 5 seats
- Included: Team conventions doc
- Not included: Reuse in client work
-
Studio
$99 $74.25
- Included: Full kit
- Included: Unlimited internal seats
- Included: Team conventions doc
- Included: Reuse in client work
Most people want Solo. Team is for standardizing three or more developers when you'd rather not answer the same question four times. Studio is for agencies putting this into repos they don't own — it comes with a one-page client-reuse grant you can hand to your client.
Refund policy
14-day no-questions refund.
Email support@agent-ops-kit.com, say "refund," done. No form, no reason required, no follow-up.
Nobody will ask you to delete anything — though the licence ends with the refund. Refunds are issued first and asked about second: if something didn't work we'd like to know so the next person doesn't hit it, but ignoring that question is a completely fine answer.