The Engineer Kit Has More Skills Than the Marketing Kit and Still Isn't Live

Two Kits, Two Different Badges

Open the “Available Kits” section on the AgentKits homepage and you’ll see two cards sitting side by side. The Marketing Kit has a green “Ready” badge, 18 agents, 93 commands, and 28 skills covering campaign planning, content, SEO, CRO, and email sequencing. Right next to it, the Engineer Kit has a yellow “Preparing” badge, 17 agents, 85 commands, and 40 skills — more skills than the kit that’s actually shipped — covering full-stack development, debugging, testing, code review, and DevOps automation.

That’s a specific, deliberate ordering worth asking about. AgentKits is an open-source, MIT-licensed toolkit built for exactly the audience that would be first in line for a coding-agent kit — developers already living inside Claude Code, Cursor, and GitHub Copilot. And yet the coding kit is the one still in progress, while a marketing-automation kit shipped first, fully built out, with room to spare on the skill count. We’ve written before about how the Marketing Kit organizes its commands by funnel stage rather than tech stack — that post mentioned the Engineer Kit only in passing. It’s worth looking at on its own.

What “Preparing” Actually Means Here

It’s worth being precise about what “Preparing” signals, because it’s not the same status as the nine other kits listed further down the same page. Sales, Legal, Finance, HR, Education, Creator, E-commerce, Healthcare, and Real Estate all sit under a separate “Coming Soon” heading, each with a much thinner spec — Sales Kit, for instance, is scoped at 6 agents and 17 commands, with no skill count published at all. Those are roadmap placeholders: a name, an icon, and a rough headcount.

The Engineer Kit isn’t that. It has a full spec — 17 agents, 85 commands, 40 skills, a public description (“Full-stack development, debugging, testing, code review, and DevOps automation”), and its own linked repository at github.com/aitytech/agentkits-engineer. It sits in the same “Available Kits” section as the shipped Marketing Kit, not down in the wishlist tier. That’s a middle state between “idea on a roadmap” and “shipped” — specced and named, actively being built, not yet at the point where the badge flips from yellow to green.

Full-Stack, But Not First

The scope described for the Engineer Kit — full-stack development, debugging, testing, code review, DevOps automation — is a broad remit, and broad is exactly the category that’s gotten crowded fastest. Industry roundups of the 2026 AI coding agent landscape name Cursor, Claude Code, Codex, GitHub Copilot, and Cline as the current front-runners, with newer entrants like Kiro drawing attention specifically around spec-driven development and DevOps automation — the same territory the Engineer Kit’s own description claims. The market itself has scaled with the crowding: the global AI code assistant market is estimated at roughly $5.42 billion in 2026, and GitHub Copilot alone has grown from 1.8 million paid subscribers in FY2024 to 4.7 million by January 2026, on top of 20 million-plus all-time users (Faros AI, “Best AI Coding Agents for 2026”; getpanto.ai, “AI Coding Statistics”). By 2026, the baseline expectation for a coding agent isn’t “autocomplete” — it’s reading a whole repo, planning a change across files, writing the code, running the tests, debugging failures, opening the PR, and answering review comments, largely unattended.

Shipping a general-purpose competitor into that field, as an eleventh entrant with less runway than Copilot or Cursor, is a genuinely hard sell. Shipping something narrower and more opinionated is a different proposition — and it’s the one the rest of AgentKits’ portfolio strategy is already built around.

Betting on Vertical, Not Volume

The Marketing Kit isn’t a general-purpose writing assistant with a marketing skin on it; it’s organized around a specific funnel and a specific set of workflows a marketing team actually runs. That’s the same design instinct the 2026 market data says is winning right now. Analyst forecasts for the vertical-agent shift are unusually sharp for this stage of the AI market: Gartner projects that task-specific AI agents will be embedded in 40% of enterprise applications by 2026, up from under 5% in 2025 — roughly an eightfold jump in a single year — and separately forecasts that 80% of enterprises will have adopted a vertical AI agent by the same point. Industry coverage of the trend frames it plainly: role-aware, workflow-specific agents are reported to complete tasks at three to five times the rate of horizontal copilots, with materially better retention, because a domain-specific agent can be precise about the tools, language, and workflow of one job rather than generically competent at everything (Lindy, “What are Vertical AI Agents?”; SaaSMag, “Vertical AI Agents Are Eating Horizontal SaaS in 2026”).

Read against that backdrop, an Engineer Kit built around full-stack development, debugging, testing, code review, and DevOps automation as named, scoped competencies — rather than a single do-everything coding chat window — isn’t a smaller ambition than “compete with Copilot.” It’s the same bet the Marketing Kit already made, applied to a second, much noisier vertical. The 40-skill count sitting ahead of the shipped kit’s 28 is a hint at the shape of that bet: more of the Engineer Kit’s value is meant to live in named, invocable skills — a debugging skill, a code-review skill, a DevOps-automation skill — rather than in general conversational capability alone.

What the Roadmap Actually Says

None of this is a claim that the Engineer Kit is finished, released, or available to use today — it isn’t, and there’s no announced date for when that changes. What’s verifiable right now is the shape of the plan: a fully shipped, funnel-organized Marketing Kit; an Engineer Kit that’s specced out in full (agents, commands, skills, its own repo) but still tagged “Preparing”; and nine more kits behind it that are name-and-headcount only. AgentKits’ own stated aggregate — 12+ kits, 100+ agents, 200+ commands, 50+ skills across the whole planned portfolio — only has one vertical finished end-to-end so far. The Engineer Kit is the next domino, walking into a coding-agent market that’s bigger and more contested than the one the Marketing Kit shipped into — but if the vertical-agent data holds, contested and undifferentiated aren’t the same problem, and going deep on one engineering workflow at a time is exactly the strategy that already worked once.