Cosmic AI Agents vs. Contentful Skills: Two Very Different Visions of AI in the CMS

On May 20, 2026, Contentful announced Contentful Skills: a free, open-source package that teaches your AI coding agent how to build on Contentful. It works with Cursor, Claude Code, GitHub Copilot, and a dozen other AI coding tools. It is genuinely well-built and solves a real problem for Contentful developers.

We're not going to dismiss it. We're going to explain why it and Cosmic AI Agents are solving fundamentally different problems, and why that difference matters for how you think about AI in your content stack.

What Contentful Skills Actually Does

Contentful Skills ships as four installable skill packages:

• contentful-guide: Core concepts and routing, helps the AI coding agent decide which skill to use.
• contentful-nextjs: SDK setup, content fetching, and Draft Mode previews for Next.js App Router and Pages Router.
• contentful-migration: Writes and runs content model migration scripts.
• contentful-personalization: Four interactive workflows covering onboarding, live debugging, health checks, and ongoing development.

The personalization skill is the flagship. It can open a real Chrome session, inspect a running page, and cross-reference it with your codebase to debug why a variant isn't rendering. That's impressive tooling.

The frame Contentful uses is accurate: "Developers don't live in documentation anymore. They live in their AI coding agent." So Contentful built a way to live there too.

What Contentful Skills does, plainly stated: it makes developers faster at building on Contentful. It is a coding assistant that understands Contentful's content model, SDK, and personalization layer. When you ask your coding agent a question about Contentful, it now has context-aware, codebase-aware answers instead of generic ones.

That is the product. It is a copilot for developers working in Contentful.

What Cosmic AI Agents Actually Do

Cosmic AI Agents are not coding assistants. They are autonomous actors that operate inside your organization's workflows, whether or not a developer is present.

Here's the concrete breakdown by agent type:

Team Agents

Team Agents live inside Slack, WhatsApp, and Telegram. They have persistent memory across conversations, respond to messages from real team members, and take action based on what they're asked. A Team Agent can receive a Slack message like "Write a blog post about our Q2 product update and publish it to the CMS" and do exactly that, without anyone opening a code editor or a CMS dashboard. They support scheduled heartbeat check-ins, meaning they can proactively surface information on a schedule, not just when asked.

Content Agents

Content Agents create, update, and manage CMS content autonomously. They run on a schedule or are triggered by webhooks. They can generate articles, update existing content objects in bulk, run content migrations, and publish or unpublish content based on rules. No developer needs to be in the loop.

Code Agents

Code Agents connect directly to GitHub repositories. They can build features, fix bugs, create pull requests, and deploy to production or preview environments, all on isolated git branches. When a Code Agent finishes its work, it opens a PR for human review. The human reviews the code; the agent did the work.

Computer Use Agents

Computer Use Agents automate browser tasks using visual AI. They can record demo videos, cross-post media between platforms, extract data from websites, and execute any workflow that requires navigating a browser, all without human intervention.

Workflows

All four agent types can be chained together using Workflows. A Workflow might look like: a Content Agent writes a blog post, a Code Agent deploys the page, and a Team Agent posts a summary to Slack, all triggered by a single webhook when a product update lands in your CMS. Workflows run on schedules or fire on events.

And all of this integrates with Cosmic's MCP Server and Agent Skills for Cursor, Claude Code, and GitHub Copilot, so developers working in AI coding environments have Cosmic context too, when they need it.

The Philosophical Difference: Copilot vs. Autonomous Operator

This is where the two products diverge at the root level.

Contentful Skills makes a developer more effective. The developer is still the actor. They type a prompt into their coding agent, the skill provides better guidance, and the developer writes better code faster. The human is in the loop at every step. This is the copilot model, and it's a good one.

Cosmic AI Agents remove the developer from the loop entirely for tasks that don't require them. A marketing manager can message an agent in Slack and get a published blog post. A content ops team can schedule an agent to monitor and update product descriptions every night. A Code Agent can respond to a bug report, fix it, open a PR, and notify the team, all while everyone sleeps.

The question isn't which approach is better in the abstract. The question is: what problem are you trying to solve?

If your problem is that Contentful is hard to set up and your developers waste time in documentation: Contentful Skills is a well-targeted solution.

If your problem is that content creation, publishing, code deployment, and distribution require too many humans doing too much manual work: you need autonomous agents, not a smarter autocomplete.

Side-by-Side Comparison

Real Capabilities, Not Marketing Claims

Let's be concrete about what Cosmic AI Agents can do right now:

A content team lead messages their Slack agent: "We need a blog post comparing headless CMS options for Next.js teams, targeting 'best headless CMS Next.js' as the primary keyword. Publish it as a draft and send me a summary."

The Team Agent researches the topic using web browsing, writes the article with proper SEO structure, creates a featured image, saves it as a draft in Cosmic, and posts a summary back to Slack, including a link to review. No developer. No dashboard login. One message.

An engineering team sets up a weekly Code Agent: Every Monday at 9am, the agent checks for open GitHub issues tagged , writes fixes on isolated branches, opens PRs with descriptions, and posts the PR links to the #engineering Slack channel for review.

A marketing team runs a multi-agent Workflow: When a new product feature ships (webhook trigger), a Content Agent drafts the announcement blog post, a Team Agent posts a teaser to the company's Slack, and a Computer Use Agent records a short demo video for the social team. All automated, all logged.

This is what "autonomous" means in practice: workflows that run without someone manually initiating each step.

On the MCP Server and Agent Skills

Cosmic does offer an MCP Server and Agent Skills for Cursor, Claude Code, and GitHub Copilot, so developers working in AI coding environments get Cosmic context when they need it. This is the overlap with Contentful Skills.

But it's a small part of the Cosmic AI story, not the center of it. The center is agents that operate on your behalf, not tools that assist you while you operate.

Pricing, Access, and Getting Started

Contentful Skills is free and open-source. Install it and your coding agent gets Contentful context. That's a genuinely good deal for Contentful developers.

Cosmic AI Agents are built into every Cosmic plan:

• Free: $0/month, includes 1 Team Agent, 2 Automations, 1 Workflow
• Builder: $49/month, includes 3 Team Agents, 5 Automations, 3 Workflows
• Team: $299/month, includes 10 Team Agents, 25 Automations, 10 Workflows
• Business: $499/month, includes 25 Team Agents, 100 Automations, 50 Workflows

The Free plan requires no credit card. You can build your first autonomous agent workflow today without spending anything.

The Bottom Line

Contentful Skills is a thoughtful developer tool. It solves a real problem for teams already on Contentful who want AI coding agents to understand their platform better. If you're in that boat, it's worth installing.

Cosmic AI Agents are a different category of product. They don't just help you build faster. They do the work: writing content, deploying code, distributing across channels, and reporting back, all without requiring a human to trigger each step.

One product accelerates a developer. The other replaces the workflow.

If you're evaluating where AI fits in your content and development stack in 2026, that distinction is worth understanding before you commit to a platform.

Ready to see what autonomous agents look like in practice? Start building free on Cosmic, no credit card required. Or book a 30-minute demo with Tony to see the full agent workflow live.