Use it the same way you would use the `FastAPI` class:

```Python hl_lines="6  11  16" title="app/routers/users.py"
{!../../docs_src/bigger_applications/app/routers/users.py!}
```

You can think of `APIRouter` as a "mini `FastAPI`" class.

All the same options are supported.

All the same `parameters`, `responses`, `dependencies`, `tags`, etc.

/// tip

        }
      } finally {
        // Important to null this out here - if we did *not* execute inline, we might still
        // run() on the same thread that called execute() - such as in a thread pool, and think
        // that it was happening inline. As a side benefit, avoids holding on to the Thread object
        // longer than necessary.
        submitting = null;
      }
    }

        // This may cause issues when an anonymous class is created inside a static initializer,
        // and the class's constructor's first parameter happens to be the enclosing class.
        // In such case, we may mistakenly think that the class is within a non-static context
        // and the first parameter is the hidden 'this'.
        return declaringClass.getEnclosingClass() != null
            && !Modifier.isStatic(declaringClass.getModifiers());

 */
@DoNotMock("Use ImmutableList.of or another implementation")
@GwtCompatible(emulated = true)
@SuppressWarnings("serial") // we're overriding default serialization
@ElementTypesAreNonnullByDefault
// TODO(kevinb): I think we should push everything down to "BaseImmutableCollection" or something,
// just to do everything we can to emphasize the "practically an interface" nature of this class.

  /*
 *
 * h2. Concerns for the plugin manager
 *
 * h3. resolving the dependencies of a plugin - these could be resolved remotely at runtime or, -
 * they could be resolved from a local repository - I think we need a simple dependency model here
 * that is more mercury related and not Maven related i.e. no POMs - workspace resolver - we need
 * tools to pre-populate this repository
 *
 * h3. create an isolated classloader
 *

             * request. However the chances are slim and the
             * retry code should ensure the overall request
             * is serviced. The alternative complicates things
             * more than I think is worth it.
             */

            if ( this.socket != null ) {
                this.socket.close();
                this.socket = null;
            }
            this.thread = null;

                    log.println( "byteCount=" + byteCount +
                            " but readBytesWireFormat returned " + n );
                }
            }
            // Don't think we can rely on n being correct here. Must use byteCount.
            // Last paragraph of section 3.13.3 eludes to this.

            bufferIndex += byteCount;
        }

        length = bufferIndex - start;

  <description>Parent for guava artifacts</description>
  <url>https://github.com/google/guava</url>
  <properties>
    <!--
    We could override this to change which version we run tests under.
    However, I think that our -Djava.security.manager=allow profile is based on the *Maven* JDK.
    If we want to use overrides here, we should change that profile to also be based on surefire.toolchain.version.
    -->

  <description>Parent for guava artifacts</description>
  <url>https://github.com/google/guava</url>
  <properties>
    <!--
    We could override this to change which version we run tests under.
    However, I think that our -Djava.security.manager=allow profile is based on the *Maven* JDK.
    If we want to use overrides here, we should change that profile to also be based on surefire.toolchain.version.
    -->

	// (e.g. LoadBalanced Service packets, etc) that need to be authenticated/captured/proxied.
	//
	// We want to do the same thing in ambient but can't rely on podSpec injection. So, do effectively the same thing,
	// but with iptables rules - use `--socket-exists` as a proxy for "is this a forwarded packet" vs "is this originating from