Bento Documentation

Bento is analytics for what your AI app does in production. This page takes you from nothing to a live trace in the dashboard: install the SDK, send one trace, watch it land.

Set up with an AI coding tool

Your AI coding tool can wire up the SDK for you. Install Bento’s Agent Skills, then paste this prompt into Claude Code, Cursor, Codex, OpenCode, or Windsurf:

Set up Bento in this project. Run npx skills add BentoLabs-ai/skills, then use the bentolabs-integrate skill for a fresh start, or bentolabs-migrate if I'm coming from another SDK. Walk me through it step by step and confirm before editing.

The skill carries the full playbook. It discovers your framework, installs the SDK, wires up bento.instrument() or bento.track_ai, and verifies the first trace lands.

Other AI coding tools

Each tool installs the skill the same way, then loads the docs over MCP. Pick yours:

Claude Code
Codex
OpenCode
Windsurf

npx skills add BentoLabs-ai/skills
claude mcp add --transport http bentolabs-docs https://docs.bentolabs.ai/mcp

Run claude in your project and paste the prompt above.

npx skills add BentoLabs-ai/skills

Add to ~/.codex/config.toml:

[mcp_servers.bentolabs-docs]
type = "http"
url = "https://docs.bentolabs.ai/mcp"

Start codex and paste the prompt above.

npx skills add BentoLabs-ai/skills

Add to opencode.json:

{
  "mcp": {
    "bentolabs-docs": {
      "type": "remote",
      "url": "https://docs.bentolabs.ai/mcp"
    }
  }
}

Run opencode and paste the prompt above.

npx skills add BentoLabs-ai/skills

In Cascade, add a custom MCP server with URL https://docs.bentolabs.ai/mcp, then paste the prompt above.

Or set it up manually

Three steps get you from an API key to a trace in the dashboard.

Get an API key

Sign in at platform.bentolabs.ai and create a workspace. Copy the API key from settings. Keys start with bl_pk_ and the prefix is validated by the SDK, so a clearly-bad key fails before any network request goes out.

export BENTOLABS_API_KEY="bl_pk_..."

Install

pip install bentolabs-sdk

Using Google ADK? Install the extra so Bento can capture every call automatically:

pip install "bentolabs-sdk[adk]"

Send your first trace

Pick the snippet that matches your stack.Google ADK. Capture every call in one line:

import bentolabs_sdk as bento

bento.init()
bento.instrument()

# Run your existing ADK agent. Bento captures it.
bento.flush()

Anything else (OpenAI, Anthropic, Bedrock, Vertex, …). Wrap each LLM call site:

import bentolabs_sdk.analytics as bento

bento.track_ai(
    event="user_message",
    user_id="user_42",
    convo_id="conv_abc",
    model="gpt-4o",
    provider="openai",
    input="What's the capital of France?",
    output="Paris.",
)
bento.flush()

Already on a framework that speaks OpenTelemetry (Vercel AI SDK, Mastra, LangChain, LlamaIndex, …)? Point its exporter at Bento instead. See Export from any framework. In TypeScript this needs no Bento package at all.Open platform.bentolabs.ai. The trace lands within seconds.

Export from any framework

Point Vercel AI SDK / Mastra / LangChain at Bento.

Integrations

Capture Google ADK with one line.

Manual tracking

Wrap individual LLM calls with bento.track_ai.

Configuration

Env vars, identity getters, lifecycle.

From Langfuse

Mechanical translation guide.

From Raindrop

Mechanical translation guide.

Overview

Migrations

Concepts

Quickstart

Set up with an AI coding tool

Other AI coding tools

Or set it up manually

Next

Export from any framework

Integrations

Manual tracking

Configuration

From Langfuse

From Raindrop

​Set up with an AI coding tool

​Other AI coding tools

​Or set it up manually

​Next

Export from any framework

Integrations

Manual tracking

Configuration

From Langfuse

From Raindrop

Set up with an AI coding tool

Other AI coding tools

Or set it up manually

Next