Skip to main content

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.

Auto-generated from autoplay_sdk/skills/. Edit the source SKILL.md and run python scripts/sync_skill_docs.py.
Download .md · Open raw file · Install via CLI: autoplay-install-skills

​
Chatbot — Botpress

Read autoplay-core first for install, credentials, and the stream wiring pattern.

​
Scoping pattern for Botpress

session_id is the table key. No conv_map needed — Botpress Tables are keyed directly by session_id. The Autonomous Agent reads the table on every message. Never mix rows from different session_id values — always filter table queries by session_id.

​
Architecture: two parallel paths

WebHook → StoreEventsData      (writes every ActionsPayload to AutoPlayEventsTable)
Start → FetchEventsData → Autonomous Agent → End   (reads table on each user message)

​
Step 1 — Create Botpress Tables

In Botpress Studio create two tables: AutoPlayEventsTable
ColumnType
titleString
descriptionString
canonical_urlString
user_idString
product_idString
session_idString
AutoPlayEventsSummaryTable
ColumnType
summaryString
session_idString

​
Step 2 — StoreEventsData code node

Webhook fires on every ActionsPayload. Write each action as a row: 
// StoreEventsData code node
const payload = event.payload;
const sessionId = payload.session_id;

for (const action of payload.actions) {
  await client.createTableRow({
    table: "AutoPlayEventsTable",
    row: {
      title: action.title,
      description: action.description,
      canonical_url: action.canonical_url ?? "",
      user_id: payload.user_id ?? "",
      product_id: payload.product_id,
      session_id: sessionId,
    },
  });
}

​
Step 3 — FetchEventsData code node

Runs before the Autonomous Agent on every user message. Fetches recent rows for this session: 
// FetchEventsData code node
const sessionId = event.userId;  // or however you surface session_id in Botpress

const rows = await client.findTableRows({
  table: "AutoPlayEventsTable",
  filter: { session_id: sessionId },
  limit: 20,
  orderBy: [{ column: "id", direction: "desc" }],
});

const summaryRows = await client.findTableRows({
  table: "AutoPlayEventsSummaryTable",
  filter: { session_id: sessionId },
  limit: 1,
});

workflow.agentContext = [
  summaryRows[0]?.summary ? `Session summary: ${summaryRows[0].summary}` : "",
  rows.rows.map(r => `- ${r.title}: ${r.description} (${r.canonical_url})`).join("\n"),
].filter(Boolean).join("\n\n");

​
Step 4 — Autonomous Agent system prompt

Add {{workflow.agentContext}} to the agent’s system prompt: 
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.

​
Step 5 — Python stream worker (posts to Botpress webhook)

import httpx
from autoplay_sdk import AsyncConnectorClient, AsyncSessionSummarizer
from autoplay_sdk.context.agent_context import AsyncAgentContextWriter

BOTPRESS_WEBHOOK_URL = "https://webhook.botpress.cloud/YOUR_WEBHOOK_ID"

async def write_actions_cb(session_id: str, text: str) -> None:
    async with httpx.AsyncClient() as http:
        await http.post(BOTPRESS_WEBHOOK_URL, json={
            "type": "actions", "session_id": session_id, "text": text
        })

async def overwrite_cb(session_id: str, summary: str) -> None:
    async with httpx.AsyncClient() as http:
        await http.post(BOTPRESS_WEBHOOK_URL, json={
            "type": "summary", "session_id": session_id, "summary": summary
        })

agent_writer = AsyncAgentContextWriter(
    summarizer=AsyncSessionSummarizer(llm=llm, threshold=20),
    write_actions=write_actions_cb,
    overwrite_with_summary=overwrite_cb,
    debounce_ms=0,
)

​
Reference