How do i make an agent with the forge?(ui_ui)

I dont know yet, lets find out(ui_ui)

Getting your API turned into an intelligent agent is surprisingly simple. All we need is your OpenAPI v3 specification and some basic auth info. That's it.

Most integration platforms require complex SDKs, extensive configuration, and ongoing maintenance. The Forge is designed to work with what you already have: your API documentation. If you've got an OpenAPI spec (and most modern APIs do by default, its literally in the framework most of the time!), you're 90% of the way there.

What We Need From You

The foundation of every Forge integration is a solid OpenAPI v3 specification. This isn't just any spec though - the quality of your AI agent directly depends on how well your spec describes your API. Think of it like this: the better you explain your API to us, the better the agent can understand it.

Your OpenAPI spec should include comprehensive descriptions for every input parameter. Don't just say "uiui_id: string" - explain what kind of user ID it is, what format it should be in, and what happens if it's invalid. The same goes for every route: describe what the endpoint actually does, not just what HTTP method it uses.

Error codes are especially important because they help the AI agent understand when something goes wrong and how to handle it. Include descriptions for all your error responses, especially the common ones like authentication failures, rate limits, and validation errors.

For authentication, we handle most standard patterns automatically. Whether you use Bearer tokens, API keys, OAuth flows, or custom auth schemes, just document it in your security schemes and we'll figure out how to integrate it properly. If you have some kind of super unique and complex auth system, it probably wont be too much of a problem, just reach out directly, we'll figure it out(ui_ui)

The Process for Dev Teams

Getting your OpenAPI spec ready for the Forge just require some attention to detail. Start by auditing your existing spec, as we said, most teams have one already, but it's often incomplete or outdated. We can work with incomplete ones, even better, we at times have bandwidth to help you make them more complete and better haha so feel free to contact us.

Go through each endpoint and make sure the descriptions actually explain what the endpoint does from a business perspective. Instead of "Updates user record", write something like "Updates user profile information including name, email, and notification preferences. Requires user to be authenticated and can only update their own profile."

Parameter descriptions should include examples and constraints. If you have an "amount" parameter, explain the currency, minimum/maximum values, and decimal precision. If you have enum values, explain what each option means in practice.

Response schemas should include sample data that represents real-world usage, not just "string" and "integer" placeholders. The AI needs to understand what your data actually looks like to generate useful insights and visualizations.

Don't forget about edge cases and error conditions. Document rate limiting behavior, pagination patterns, and any quirks in your API that developers typically discover the hard way. The more complete your spec, the more intelligent your resulting agent will be.

Again, this is more of some general guidelines, i hope if youve gotten to this stage, you know how to write fucking api docs lol. or just ask claude.

Getting Started

The easiest way to get started is to reach out to us directly through our Telegram channel. We're not doing automated onboarding yet because every API integration benefits from some human review to make sure we're maximizing the potential of your service.

When you contact us, have your OpenAPI spec ready to share. We'll review it, give you feedback on any improvements that would make your agent more powerful, and walk you through the integration process. The whole thing usually takes a few hours(sometimes minutes) from first contact to having a working AI agent.

We're currently doing free initial integration demos for teams that want to explore how the Forge can transform their APIs. This isn't a sales pitch, it's a technical collaboration to make sure your API becomes as intelligent and useful as possible.