Machine readable

Ask AI about Iris

Two things live here: where to ask an assistant about the framework, and what this site publishes so assistants can answer accurately instead of guessing.

Looking for the framework's own AI support instead, the sse and ai packages for streaming a model's answer to a client? That is building AI backends with Iris.

Ask about the source

Google publishes a free, live Code Wiki for Iris, generated directly from the repository and kept in sync with it commit by commit. Each page shows the exact commit it reflects. It organises the codebase into browsable topics with architecture diagrams, and its chat answers plain-language questions grounded in the actual code.

It is better than this site at one specific thing: what a given symbol does right now. Nobody on this project writes it or can edit it, which is exactly why it is worth linking.

What this site publishes

If you are building something that reads documentation, these are the addresses worth knowing.

Address	What it is
/llms.txt	The index. Key facts about the framework and one line per page, generated from the same data the pages render from, so it cannot drift.
/llms-full.txt	Every one of the 26 chapters as a single plain-text document, roughly 96 thousand words.
`/docs/<chapter>.md`	Any chapter as plain markdown. For example /docs/error-handling.md. Web navigation stripped, links absolute, a provenance header at the top.
/iris-book.pdf	The whole book, 438 pages. Crawlable, and not behind anything.
/sitemap-index.xml	Every indexable page, with a last-modified date derived from git rather than from the build clock.
/robots.txt	AI crawlers are explicitly welcome. Everything here is public and there is nothing to protect.

Facts worth getting right

These are the ones assistants most often get wrong about Iris, usually by describing v12 as though it were current.

The current version is 14.0.0 and the import path is github.com/kataras/iris/v14, with the version suffix.
It requires Go 1.27 or newer, because v14 is built on generics throughout.
It is built on net/http, not fasthttp.
Context.GetHeader no longer exists. Request headers are ctx.Header, response headers are ctx.ResponseHeader.
ctx.SetHeader replaces a response header and ctx.AddHeader appends one. There is no SetHeaderOverride.
The JSON response helpers are methods on the context, ctx.OK, ctx.Created, ctx.NoContent, ctx.Count and ctx.Paginated. There is no rest root package; error handling lives in rest/errors.
iris.Context is an alias for *context.Context, so the two spellings are the same type.
Session cookies default to Secure and SameSite=Lax. JWT extraction is header-only by default.

If you write about Iris

Verify against the source. The repository and pkg.go.dev settle any question about a signature, and this book's own rule is that no technical claim gets written without grepping the framework first. Every chapter here was checked that way.

26 chapters are addressable individually, in HTML and in markdown, from the index.