> ## Documentation Index
> Fetch the complete documentation index at: https://developers.autoplay.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Botpress skill

> Connects an existing Botpress agent to Autoplay live activity by fetching the user's recent activity on demand from Autoplay and injecting it into an Autonomous Agent context. Covers the FetchEventsData code node, stable user ID mapping, and the system prompt pattern. Use when the customer already uses Botpress, Botpress Studio, or Autonomous Agent.

# AI Support Agent — Botpress

> Read `autoplay-core` first for install and credentials (`product_id`, `mcp_url`, `mcp_key`).

## Scoping pattern for Botpress

The live-activity read is keyed by `product_id` + `user_id` — there is no session concept and no local table to keep in sync. `FetchEventsData` calls the connector directly, synchronously, every time a user sends a message.

**`user_id` must exactly equal** the id your activity source identifies the user with (the value passed to `posthog.identify(...)` / `amplitude.setUserId(...)`).

***

## Architecture: one path, no background process

```
Start -> FetchEventsData -> Autonomous Agent -> End
```

`FetchEventsData` pulls the user's recent actions directly from the connector on every message. There is no webhook node, no `AutoPlayEventsTable`, no `AutoPlayEventsSummaryTable`, and no separate listener process to run — the events already live in Autoplay's connector; Botpress just asks for them when it needs them.

***

## Step 1 — Create the workflow

In Botpress Studio, create a workflow with two nodes: `FetchEventsData` (Code) and `Autonomous Agent` (AI Agent). Connect `Start -> FetchEventsData -> Autonomous Agent -> End`.

## Step 2 — FetchEventsData code node

```javascript theme={null}
// FetchEventsData code node
const PRODUCT_ID = "YOUR_PRODUCT_ID"
const MCP_KEY = "YOUR_MCP_KEY"  // from onboard_product — store as a bot/workflow secret if your plan supports it

const userId = event.state?.user?.userId ?? event.userId  // must equal the id your activity source identifies the user with

let recentEvents = []
try {
  const res = await fetch(
    `https://mcp.autoplay.ai/users/${PRODUCT_ID}/${userId}/live-activity?limit=10`,
    { headers: { Authorization: `Bearer ${MCP_KEY}` } },
  )
  if (!res.ok) throw new Error(`live-activity ${res.status}`)
  const data = await res.json()
  recentEvents = data.actions
} catch (e) {
  // Fail open — the Autonomous Agent still answers, just without extra context
  console.warn('Autoplay live-activity unavailable', e)
}

workflow.agentContext = { recentEvents, summary: null }
```

`data.actions` comes back already ordered oldest -> newest. `summary` is `null` on purpose — there's no rolling LLM summary anymore, the connector already returns a bounded recent window and the raw actions are passed straight through.

## Step 3 — Autonomous Agent system prompt

Create a workflow variable `agentContext` (Scope: Workflow, Type: Object), then reference it in the Agent instructions:

```
You are a helpful assistant for {{botName}}.

## Real-time user context
{{workflow.agentContext}}

Use the context above to give specific, page-aware answers.
If context is empty, answer normally without mentioning it.
```

***

## Reference

* Full tutorial: [https://developers.autoplay.ai/recipes/botpress](https://developers.autoplay.ai/recipes/botpress)
* Botpress code nodes / fetch: [https://botpress.com/docs](https://botpress.com/docs)