OpsKit vs AgentsKit
AgentsKit gives Claude Code 89 generalist agents across engineering and marketing. OpsKit gives it a read-only connection to your live Kubernetes cluster. A side-by-side on scope, safety, and price.
Facts about AgentsKit taken from agentskit.co, checked 29 July 2026. Anything their site doesn't state is marked "not stated" rather than guessed at.
What AgentsKit is
A Claude Code kit of 89 specialist agents, 103 skills and 181 slash commands spanning engineering and marketing roles, sold as a one-time purchase.
What OpsKit is
A Kubernetes and cloud investigation kit for AI coding agents. It adds a read-only MCP connection to your live cluster, investigation skills with explicit evidence requirements, and a safety layer that blocks every mutation verb before it executes.
The short version
These solve different problems. AgentsKit is breadth — a generalist team of prompts covering everything from React to SEO. OpsKit is depth on one thing: it connects your agent to a real cluster so an incident can be diagnosed from evidence instead of from training data. If you want one kit to cover your whole product, AgentsKit covers far more ground. If your problem is that your agent guesses about production, prompts alone won't fix it.
Side by side
| Dimension | OpsKit | AgentsKit |
|---|---|---|
| What it is | A Kubernetes and cloud investigation kit with a live cluster connection | A prompt kit of generalist engineering and marketing agents |
| Contents | 42 agents, 33 skills, 47 commands + a 15-tool MCP server | 89 agents, 103 skills, 181 slash commands |
| Reads your live cluster | Yes — 15 read-only MCP tools over your kubeconfig | Not stated on their site |
| Mutation safety | Allowlist hook blocks mutations at the OS level (opskit-guard, MIT) | No safety or read-only guarantee stated |
| Scope | Kubernetes, AWS, GCP, Azure, Terraform, SRE and incident response | Full-stack engineering plus marketing, growth, SEO and brand |
| Agent support | Claude Code, Cursor, Codex CLI, Kiro and other MCP agents | Claude Code only |
| Price | $19 one-time, launch price (list $99) | $49 per kit, $89 for the bundle — one-time |
Choose AgentsKit if…
- You want one kit that covers your whole product surface — frontend, backend, database, plus marketing and SEO.
- You're a solo builder or small team, and generalist coverage matters more than depth in any single domain.
- You don't run Kubernetes, so a live cluster connection buys you nothing.
- You want the largest raw agent count for the money.
Choose OpsKit if…
- Your agent needs to read the actual cluster — pod events, previous-container logs, resource state — not answer from training data.
- You need an enforced guarantee that the agent cannot run kubectl apply, not a prompt asking it not to.
- You want fixes to arrive as reviewable GitOps PRs with the evidence attached.
- You use Cursor, Codex CLI or Kiro alongside Claude Code.
Questions
Can I use both together?
+
Yes. They install into the same .claude/ directory and neither replaces the other. AgentsKit's agents cover application work; OpsKit's cover the cluster and the cloud accounts around it. The safety hook OpsKit installs applies to every bash command in the project, whichever agent issued it.
Is a bigger agent count better?
+
Not by itself. Agent count measures how many prompt files ship, not how much either kit can see. OpsKit's differentiator is the read-only MCP server, which is why it ships fewer agents and a live cluster connection instead.
Try the free tool first
The Kubernetes YAML linter runs the same class of checks OpsKit's deployment guard applies — free, in your browser, no signup. It's the cheapest way to see how the kit thinks.