docs: prep for site

draft.md

@@ -0,0 +1,308 @@
+# RivetKit Setup & Deploy
+
+Read https://rivet.dev/llms.txt to understand how RivetKit works (actors, state, events, actions, connections, clients).
+
+Add a note to the project's \`CLAUDE.md\` / \`AGENTS.md\` (create one if missing) linking to https://rivet.dev/llms.txt as the reference for future RivetKit work.
+
+## Step 1: New project or existing?
+
+Ask the user:
+
+> Do you want to **build a new project with RivetKit**, or **integrate RivetKit into this existing project**?
+
+Branch based on their answer.
+
+---
+
+## Path A: New project
+
+### 1. Suggest something to build
+
+Recommend something realtime, fast to spin up, and with a simple frontend. Good starting points:
+
+- **Chat app** (rooms + messages, great default)
+- Collaborative todo / list
+- Live cursors / presence demo
+- Simple multiplayer game (tic-tac-toe, drawing board)
+
+Ask what they want to build. If they're unsure, default to a chat app.
+
+### 2. Build it
+
+Scaffold a minimal project with RivetKit:
+
+- \`npm install rivetkit\` (or pnpm/yarn/whatever is being used)
+- Add a frontend (plain HTML/JS or React via \`@rivetkit/react\` — keep it small).
+- Define actors + registry (see https://rivet.dev/docs/actors).
+- Serve via \`registry.listen({ port: 3001, publicDir: "<frontend-output-dir>" })\` so one command serves both API and frontend.
+- Add a local dev script (e.g. \`npm run dev\`) that builds the frontend and starts the server.
+
+Reference quickstarts:
+- Node.js / Bun: https://rivet.dev/docs/actors/quickstart/backend
+- React: https://rivet.dev/docs/actors/quickstart/react
+
+### 3. Verify
+
+- \`GET /api/rivet/metadata\` returns 200.
+- Drive actors via the inspector HTTP API: https://rivet.dev/docs/actors/debugging
+- If there's a frontend, open it in a browser and confirm the realtime behavior works end-to-end.
+
+### 4. Wrap up
+
+Give the user:
+
+- A brief overview of the project (what was built, where the actors live, how state/events flow).
+- The command to run locally.
+- Next steps to deploy + Rivet link.
+
+**Deploy steps:**
+
+1. Review the docs for ${provideStr} (the following docs are available, pick the one that relates to ${provideStr}
+    https://rivet.dev/docs/connect/rivet-compute/
+    https://rivet.dev/docs/connect/aws-ecs/
+    https://rivet.dev/docs/connect/aws-lambda/
+    https://rivet.dev/docs/connect/cloudflare/
+    https://rivet.dev/docs/connect/custom/
+    https://rivet.dev/docs/connect/freestyle/
+    https://rivet.dev/docs/connect/gcp-cloud-run/
+    https://rivet.dev/docs/connect/hetzner/
+    https://rivet.dev/docs/connect/kubernetes/
+    https://rivet.dev/docs/connect/railway/
+    https://rivet.dev/docs/connect/supabase/
+    https://rivet.dev/docs/connect/vercel/
+    https://rivet.dev/docs/connect/vm-and-bare-metal/
+)
+2. Configure and deploy using the following environment variables:
+  RIVET_PUBLIC_ENDPOINT=${publishableToken}
+  RIVET_ENDPOINT=${secretToken}${
+		runnerName !== "default"
+			? \`
+  RIVET_POOL=${runnerName}\`
+			: ""
+  }
+3. Finally, if applicable (not necessary for Rivet Compute), tell the user the URL to paste in to the Rivet dashboard
+
+---
+
+## Path B: Integrate into existing project
+
+### 1. Understand the project + propose
+
+Read through the project to understand what it does. Then ask:
+
+> What do you want to build, or how do you want to integrate RivetKit?
+
+Give recommendations grounded in what you see. Examples:
+- Chat app with video → move chat into a RivetKit actor for realtime + state.
+- Todo app → per-list actor for collaborative editing.
+- Dashboard → per-user actor for live metrics / presence.
+
+### 2. Plan
+
+Describe the plan before writing code:
+- Which features become actors.
+- What state each actor holds.
+- Which actions + events the frontend will use.
+
+Get confirmation before implementing.
+
+### 3. Integrate (actors only)
+
+- \`npm install rivetkit\`
+- Add actors + registry (see https://rivet.dev/docs/actors).
+- Mount on the existing server: \`app.all("/api/rivet/*", (c) => registry.handler(c.req.raw))\` (or the equivalent for the project's framework).
+- Do **not** touch the frontend yet unless the user asks.
+
+Verify with \`/api/rivet/metadata\` and the inspector API (https://rivet.dev/docs/actors/debugging).
+
+### 4. Wrap up
+
+Give the user:
+- A short overview of what was added.
+- The command to run locally.
+
+Then ask:
+
+1. **Want to integrate this into the frontend?** Point at https://rivet.dev/docs/clients/react (or the relevant client doc) and wire it up if yes.
+2. **Want to deploy?** If yes, **Deploy steps:**
+
+    a. Review the docs for ${provideStr} (the following docs are available, pick the one that relates to ${provideStr}
+        https://rivet.dev/docs/connect/rivet-compute/
+        https://rivet.dev/docs/connect/aws-ecs/
+        https://rivet.dev/docs/connect/aws-lambda/
+        https://rivet.dev/docs/connect/cloudflare/
+        https://rivet.dev/docs/connect/custom/
+        https://rivet.dev/docs/connect/freestyle/
+        https://rivet.dev/docs/connect/gcp-cloud-run/
+        https://rivet.dev/docs/connect/hetzner/
+        https://rivet.dev/docs/connect/kubernetes/
+        https://rivet.dev/docs/connect/railway/
+        https://rivet.dev/docs/connect/supabase/
+        https://rivet.dev/docs/connect/vercel/
+        https://rivet.dev/docs/connect/vm-and-bare-metal/
+    )
+    b. Configure and deploy using the following environment variables:
+        RIVET_PUBLIC_ENDPOINT=${publishableToken}
+        RIVET_ENDPOINT=${secretToken}${
+                runnerName !== "default"
+                    ? \`
+        RIVET_POOL=${runnerName}\`
+                    : ""
+    }
+    c. Finally, if applicable (not necessary for Rivet Compute), tell the user the URL to paste in to the Rivet dashboard
+
+Link docs:
+- Actors: https://rivet.dev/docs/actors
+- Clients: https://rivet.dev/docs/clients
+- Troubleshooting: https://rivet.dev/docs/actors/troubleshooting
+
+---
+
+## If you get stuck
+
+Check https://rivet.dev/docs/actors/troubleshooting. If that doesn't help, point the user at:
+- Discord: https://rivet.dev/discord
+- GitHub issues: https://github.com/rivet-dev/rivet
+
+Include in the report: symptoms, what was tried, RivetKit version, runtime, HTTP router.
+
+
+----
+## Step 1: Load Rivet Skill
+
+Load the Rivet skill. If the skill is not installed, tell the user to run this and restart the agent:
+
+\`\`\`
+npx skills add rivet-dev/skills -s rivetkit
+\`\`\`
+
+## Step 2: Create Dockerfile
+
+If the project does not already have a \`Dockerfile\`, create one. Use this as a starting point and adjust the package manager (npm/pnpm/yarn), file paths, and entrypoint to match the project:
+
+\`\`\`dockerfile
+FROM node:24-alpine
+
+WORKDIR /app
+
+COPY package.json package-lock.json ./
+RUN npm ci
+
+COPY . .
+RUN npm run build --if-present
+
+CMD ["node", "dist/index.js"]
+\`\`\`
+
+If the project does not already have a \`.dockerignore\`, create one:
+
+\`\`\`
+node_modules/
+dist/
+.env
+.git/
+\`\`\`
+
+If Docker is installed, build and run the image to verify it works before proceeding:
+
+\`\`\`bash
+docker build -t rivet-test . && docker run --rm -p 3000:3000 rivet-test
+\`\`\`
+
+Verify the container starts and is connectable (e.g. \`curl http://localhost:3000/health\`). If Docker is not installed, skip this and proceed.
+
+## Step 3: Add GitHub Secret
+
+Add the secret RIVET_CLOUD_TOKEN to your GitHub repository:
+
+\`\`\`
+gh secret set RIVET_CLOUD_TOKEN --body "${cloudToken}"
+\`\`\`
+
+## Step 4: Create GitHub Action
+
+Create .github/workflows/rivet-deploy.yml using rivet-dev/deploy-action@v1.1:
+
+\`\`\`
+name: Rivet Deploy
+
+on:
+  pull_request:
+    types: [opened, synchronize, reopened, closed]
+  push:
+    branches: [main]
+  workflow_dispatch:
+
+concurrency:
+  group: rivet-deploy-\${{ github.event.pull_request.number || github.ref }}
+  cancel-in-progress: true
+
+jobs:
+  rivet-deploy:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      pull-requests: write
+    steps:
+      - uses: actions/checkout@v4
+      - uses: rivet-dev/deploy-action@v1.1
+        with:
+          rivet-token: \${{ secrets.RIVET_CLOUD_TOKEN }}
+\`\`\`
+
+## Step 5: Push and Verify
+
+Push the changes and monitor the GitHub Action until it completes:
+
+1. Push to the repository.
+2. Use \`gh run list\` and \`gh run watch\` to poll the action status. Do not move on until the run finishes.
+3. If the action succeeds, proceed to Step 6.
+4. If the action fails, inspect the logs with \`gh run view --log-failed\`, fix the issue, push again, and repeat from step 2.
+
+## Step 6: Verify Deployment
+
+**Token types used in this step:**
+- \`cloud_api_*\` (the \`RIVET_CLOUD_TOKEN\` secret) — a management token scoped to the Cloud API (cloud-api.rivet.dev). Use this for admin operations like checking deployment status and fetching logs.
+- \`pk_*\` (the publishable token below) — a public key scoped to the Rivet Engine API (api.rivet.dev). Use this for creating actors and calling gateway endpoints.
+
+These are different tokens with different scopes. Do not mix them up.
+
+Once deployed, verify the deployment works:
+
+1. Poll the deployment status every 5 seconds until status is "ready". Stop and investigate if status is "error".
+   \`\`\`bash
+   curl -s "${cloudApiUrl}/projects/${project}/namespaces/${cloudNamespace}/managed-pools/default?org=${organization}" \\
+     -H "Authorization: Bearer ${cloudToken}"
+   \`\`\`
+
+2. Create an actor. Actors require a key field (string, not array):
+   \`\`\`bash
+   curl -X POST "${apiUrl}/actors?namespace=${namespace}" \\
+     -H "Authorization: Bearer ${publishableToken}" \\
+     -H "Content-Type: application/json" \\
+     -d '{"name": "<ACTOR_NAME>", "key": "<KEY>", "runner_name_selector": "default", "crash_policy": "restart"}'
+   \`\`\`
+   Replace \`<ACTOR_NAME>\` with a valid actor name from the registry and \`<KEY>\` with an appropriate key string (e.g. "general"). Note the \`actor_id\` from the response.
+
+3. Wait ~10 seconds for the actor to start, then hit its health endpoint through the gateway using the public token:
+   \`\`\`bash
+   curl "${apiUrl}/gateway/<ACTOR_ID>/health" \\
+     -H "x-rivet-token: ${publishableToken}"
+   \`\`\`
+   This should return ok with a 200 status.
+
+4. If the health check returns actor_runner_failed, check the runner logs via SSE to diagnose:
+   \`\`\`bash
+   curl --max-time 15 "${cloudApiUrl}/projects/${project}/namespaces/${cloudNamespace}/managed-pools/default/logs?org=${organization}" \\
+     -H "Authorization: Bearer ${cloudToken}"
+   \`\`\`
+
+5. Common issues:
+   - "actor should have a key": The key field was missing from the create request.
+   - Token 401: Make sure you're using the correct API URLs (${apiUrl}, ${cloudApiUrl}).
+
+## Troubleshooting
+
+- There is no Rivet CLI. Do not attempt to use or install one. All deployment is done via the GitHub Action and all interaction is done via HTTP APIs (curl).
+- Architecture: The GitHub Action builds your Docker image and pushes it to Rivet. Rivet runs the container serverlessly. When you create an actor, Rivet communicates with the \`/api/rivet/*\` endpoint inside the container to manage its lifecycle.
+- For more troubleshooting help, see: https://rivet.dev/docs/actors/troubleshooting/

rivetkit-typescript/packages/rivetkit/src/registry/index.ts

@@ -1,6 +1,8 @@
+import { Hono } from "hono";
 import { ENGINE_ENDPOINT } from "@/common/engine";
 import { configureServerlessPool } from "@/serverless/configure";
-import { VERSION } from "@/utils";
+import { detectRuntime, VERSION } from "@/utils";
+import { crossPlatformServe, loadRuntimeServeStatic } from "@/utils/serve";
 import {
 	type RegistryActors,
 	type RegistryConfig,
@@ -279,6 +281,36 @@ export class Registry<A extends RegistryActors> {
 		};
 	}
 
+	/**
+	 * Starts an HTTP server that dispatches every request through the
+	 * serverless handler. Uses `crossPlatformServe` to pick the right
+	 * runtime (Node, Bun, Deno).
+	 *
+	 * @param opts.port      Port to listen on. Defaults to 3000.
+	 * @param opts.publicDir If set, serves static files from this directory
+	 *                       before falling through to the registry handler.
+	 *
+	 * @example
+	 * ```ts
+	 * await registry.listen();
+	 * await registry.listen({ port: 8080, publicDir: "./public" });
+	 * ```
+	 */
+	public async listen(
+		opts: { port?: number; publicDir?: string } = {},
+	): Promise<void> {
+		const port = opts.port ?? 3000;
+		const config = this.parseConfig();
+		const runtime = detectRuntime();
+		const app = new Hono();
+		if (opts.publicDir) {
+			const serveStatic = await loadRuntimeServeStatic(runtime);
+			app.use("*", serveStatic({ root: opts.publicDir }));
+		}
+		app.all("*", (c) => this.handler(c.req.raw));
+		await crossPlatformServe(config, port, app, runtime);
+	}
+
 	public async diagnostics(): Promise<RegistryDiagnostics> {
 		const candidates = [
 			this.#runtimeServerlessPromise,

website/src/content/docs/connect/rivet-compute.mdx

@@ -111,6 +111,22 @@ The dashboard shows live status as Rivet Compute provisions your backend:
 
 Once the status reaches **Ready**, your backend is live and actors are available for connections.
 
+<LLMOnly>
+
+If you are an agent monitoring the deployment via API rather than the dashboard, poll the managed-pool endpoint on the Cloud API.
+
+The `RIVET_CLOUD_TOKEN` secret is a `cloud_api_*` management token scoped to the Cloud API at `cloud-api.rivet.dev`. Use it for `Authorization: Bearer ...` against the Cloud API. Do not confuse it with a `pk_*` publishable key, which is scoped to the Rivet Engine API at `api.rivet.dev` and will 401 against this endpoint.
+
+Substitute `$CLOUD_API_URL` (typically `https://cloud-api.rivet.dev`), `$PROJECT`, `$ORG`, `$CLOUD_NAMESPACE`, and `$CLOUD_TOKEN`.
+
+Poll every 5 seconds until `status` is `ready`. Stop and investigate if `status` is `error`.
+
+```bash
+curl -s "$CLOUD_API_URL/projects/$PROJECT/namespaces/$CLOUD_NAMESPACE/managed-pools/default?org=$ORG" -H "Authorization: Bearer $CLOUD_TOKEN"
+```
+
+</LLMOnly>
+
 </Step>
 </Steps>