See it in action

The fastest way to understand what Forge does is to watch it run.

This demo shows Claude Desktop connected to a live Forge instance. One session. One AI agent. The full content lifecycle from start to finish.

A conversation, not a workflow

You type a sentence. The agent creates a draft, writes content that reflects your vision, and confirms what it did. You ask it to publish. It publishes. You stay in the conversation -- you do not switch tools, fill out forms, or manage state yourself.

That is what the demo shows first. The lifecycle enforcement is what makes it possible to work that way -- you can hand the agent real access because the rules hold, and the agent knows it.

What you are watching

The agent creates a draft post. The post is not visible anywhere -- not in the feed, not in the sitemap, not to search engines or AI indexes. Draft means draft.

The agent publishes it. The site updates. Sitemap, RSS feed, and AI index reflect the change immediately.

The agent archives it. The post disappears from the listing. The URL still resolves -- archived content does not vanish, it steps back.

Then the agent is asked to delete it permanently.

It refuses.

> Permanent deletion is an action I can't perform -- you'll need to do that one yourself.

That line is not a configuration. It is not a prompt instruction. It is the framework enforcing a boundary the agent cannot cross, regardless of what it is asked to do.

What this means for your team

You are not trusting the agent to behave. You are trusting the rules that govern what the agent is allowed to do -- the same rules that govern your team. The agent operates within them because it has no other option.

Read about content lifecycle · Read about agent access rules · Get started with Forge