Why put a model behind a slash command
Teams already live in Slack, so a slash command is often the cheapest way to put a model where the questions are. The hard part is not the Slack wiring, it is the model access: one key that reaches every provider, retries when one upstream fails, and a bill you can read.
AnyRouter sits between Slack and the providers, so the bot sends one request to one base URL and the gateway handles the rest.
What you need
- A Slack app with a bot token, a signing secret, and a slash command called /anyr.
- An AnyRouter API key from the dashboard. It starts with sk-ar-v1-.
- Node 20 or newer, and the Bolt framework.
Create the Slack app at https://api.slack.com/apps, add the slash command, install it to your workspace, and copy the two credentials.
The handler
Ack the command first, then call the OpenAI-compatible chat endpoint. The model defaults to anyrouter/free, which picks the best free model available:
import { App } from "@slack/bolt";
const app = new App({
token: process.env.SLACK_BOT_TOKEN,
signingSecret: process.env.SLACK_SIGNING_SECRET,
});
app.command("/anyr", async ({ command, ack, say }) => {
await ack();
const prompt = command.text?.trim();
if (!prompt) {
await say("Usage: /anyr <your question>");
return;
}
const response = await fetch("https://anyrouter.dev/api/v1/chat/completions", {
method: "POST",
headers: {
"content-type": "application/json",
authorization: `Bearer ${process.env.ANYROUTER_API_KEY}`,
},
body: JSON.stringify({
model: "anyrouter/free",
messages: [{ role: "user", content: prompt }],
}),
});
const body = await response.json();
await say(body.choices[0].message.content);
});
await app.start(Number(process.env.PORT ?? 3000));A working, tested version of this bot lives in the AnyRouterHQ/anyrouter-slack-bot repo.
Get the demo repoRun it
Install, export the three values, and start the app:
npm install @slack/bolt
export SLACK_BOT_TOKEN=xoxb-...
export SLACK_SIGNING_SECRET=...
export ANYROUTER_API_KEY=sk-ar-v1-...
npm startFor local development, point the slash command at a tunnel to your machine. In production, deploy the same process behind the public URL Slack expects.
Where to take it next
The same pattern works for a channel assistant, a support triage bot, or an on-call helper. Pin a model per command with the model field, add a usage log, and keep the key on the server so it never reaches a browser.
Free models are $0 per token on Go and up, with no daily cap, which makes a team bot cheap to run. Bring your own provider keys and the markup stays $0.
Pick a plan and start
Create an account, choose Go, Pro, Pro+, or Max, pay on the next screen, then call the model.