prompt2botBuilding agent infrastructure from scratch is a pain. You end up wrestling with context windows, conversation memory, viewers, streaming, tool routing, and channel adapters. prompt2bot provides all the agent infrastructure out of the box. You just bring your API.
Wiring a model call to an API endpoint sounds straightforward on day one. Then reality hits.
You have to manage conversation context so turns do not blow past model token limits. You need persistent memory so the agent remembers earlier requests across days. You need conversation viewers and debug traces to figure out why an agent called the wrong parameter in production. And you have to write custom adapters for web widgets, WhatsApp, and Telegram.
prompt2bot solves the entire infrastructure layer. Your system stays a regular API. prompt2bot provides the multi-turn agent runtime, the history store, the conversation viewers, and the channel gateways.
Give your coding agent our llms.txt and your API token. It handles the rest.
Generate an API token from your Settings page. This token lets your coding agent manage bots, register tools, and configure prompts programmatically.
Give Claude Code, Cursor, OpenCode, or Aider our llms.txt file along with your API docs. It reads the client spec and maps your endpoints to prompt2bot tools.
Your agent is immediately live. Drop the end-to-end encrypted chat widget into your web app, or connect your WhatsApp and Telegram channels in one click.
Paste this into Claude Code, Cursor, OpenCode, or Aider:
Read https://prompt2bot.com/llms.txt for the API reference.
My prompt2bot API token: <YOUR_P2B_TOKEN>
My system API endpoints and docs: <YOUR_API_ENDPOINTS_OR_OPENAPI_URL>
Using the @prompt2bot/client SDK or the prompt2bot HTTP API:
1. Create a bot using createBotApi.
2. Register my API endpoints as remote tools using setCustomTools with typed parameter schemas.
3. Set the system prompt so the agent validates user input, calls the right endpoints, and asks for confirmation before executing sensitive mutations.One API connection powers every channel. No separate SDKs or protocols.
Embed a private, secure chat experience directly in your web application.
Let users interact with your system on the chat apps they use all day.
Everything between your API and the user is managed by prompt2bot.
No manual prompt packing or trimming. prompt2bot tracks session limits and handles dynamic summarization automatically.
Immutable, append-only event streams in Deno KV. Conversations persist across turns, devices, and sessions.
Full monitoring out of the box with View Chat. Inspect exact model thoughts, tool parameters, execution latency, and token costs.
Tool requests to your backend are signed with SHA-256 using your bot Webhook Secret. Zero forged or replayed requests.
No custom connectors for WhatsApp Cloud API, Telegram Bot API, Slack events, or web sockets. It is all pre-wired.
Gate irreversible API calls with 1-tap confirmation. Admins approve or deny operations directly from WhatsApp or Slack.
Every tool call sent to your API is cryptographically signed using your bot Webhook Secret.
// Your API receives cryptographically signed tool requests:
POST https://your-api.example.com/tools/orders
Headers:
x-signature: <HMAC-SHA256 signature>
x-timestamp: 1715000000000
Body:
{
"orderId": "ord_9821",
"action": "refund",
"amount": 15.00,
"reason": "shipping_delay"
}Grab an API token, give prompt2bot.com/llms.txt to your coding agent, and deploy your agent across Web, WhatsApp, and Telegram in minutes.