Skip to main content
OneShot gives the agent you already run real-world actions: research, email, voice, web, and commerce. It executes each call, the SDK pays for it, and every paid action comes back with a signed receipt. Your model, framework, and hosting stay where they are.

Add your first tool

Install the SDK, connect a funded wallet, and make your first call.

Explore the tools

Find the action your agent needs, with its SDK method and API reference.

Two Layers

The SDK acts. The dashboard shows what each action cost and what it delivered.
  1. OneShot Agent SDK (execution). Real-world tools for your agent, with payment and a receipt for each paid action.
  2. Commerce Dashboard (visibility and control). Track spend by agent, set budgets, and measure return on spend.

Key Features

email

Email & Voice

Send email and SMS and make phone calls. OneShot provisions domains and registers phone numbers for you.
research

Deep Research

Research reports with source citations.
person

Person Intelligence

Dossiers, social profiles, articles, interests, and social graph mapping for any person.
commerce

Commerce

Search and purchase products from major retailers.
enrichment

Data Enrichment

Enrich profiles, find emails, verify deliverability.
web

Web Actions

Search the web, read pages, and run browser tasks.

How It Works

You call a tool. OneShot runs it and returns the result.
  1. Call a tool through the SDK with its inputs.
  2. OneShot runs it, with error recovery and retries through completion. The SDK handles quotes, payment, and job polling. Wallet sessions sign USDC payments on Base; access-token sessions spend available credits without wallet signing.
  3. Use the result in your next step. Your agent’s logic stays in your existing stack.
Use a Coinbase CDP wallet or bring your own key. Follow the quickstart to connect and fund your wallet.

Guaranteed Execution

Once OneShot accepts an action, it owns execution, provider recovery, and retries, within your spending limits. You supply valid inputs, permissions, and funding. The commitment covers the tool action (sending an email, for example), not the recipient’s response. If a job returns failed, check job status and contact support with the request ID before retrying. The action may already have executed.

Commerce Dashboard

Every paid action leaves a signed receipt, tagged by tool and agent. Your agent spends your money. OneShot writes the expense report. Track costs and results by agent, inspect receipts, and set spending limits.
receipt

Signed Receipts

Each completed or failed paid action has a signed receipt. Verify it yourself with the SDK’s verifyReceipt() or the oneshot-verify-receipt CLI.
spend

Categorized Spend

Each call is tagged by tool and agent as it happens, instead of arriving as a monthly total.
Set limits with spend budgets. Attach the reason for a call with audit context.

Philosophy

OneShot is built for one-shot interactions:
  • Each API call is independent and self-funded
  • Calls need minimal context
  • Responses are clean and deterministic
  • Each action gets a receipt
Outputs favor machine-readability and determinism over human-friendly text that is hard to parse.

Choose Your Integration

Pick the integration that matches your stack.

MCP Server

Use OneShot in Claude Desktop, Cursor, or Claude Code

TypeScript SDK

npm install @oneshot-agent/sdk

Python SDK (LangChain)

pip install langchain-oneshot. Every tool as a LangChain BaseTool

Python SDK (Core)

pip install oneshot-python: direct HTTP client with x402

Virtuals GAME SDK

Give Virtuals Protocol agents real-world tools

SDK Examples

Complete code examples

API Reference

All available endpoints

Machine-readable spec

The full API is an OpenAPI 3.1 specification: 82 operations across 19 tag groups, regenerated on every deploy from the Zod schemas the server uses at runtime. Use it to generate clients, register with API directories, or import into Postman, Insomnia, ChatGPT actions, or Claude / OpenAI assistants. The OpenAPI Specification page has SDK generation examples.

Soul.Markets

Soul.Markets measures agents with blinded evaluations, public leaderboards, and quality tiers, so agents are scored on their work.

Soul.Markets

The benchmark for agent services

Seller Guide

How to list, get measured, and earn