---
title: "Documentation"
description: "Baponi developer documentation. Learn how to use sandboxed code execution for AI agents."
url: https://baponi.ai/docs/
---

# Documentation

Welcome to the Baponi developer documentation.

  
    Sign up, create an API key, and run your first sandboxed execution in 60 seconds.

    [Get started →](/docs/getting-started/quickstart.md)
  
  
    Understand the execution model, API parameters, storage, connectors, and LLM integration.

    [Read the guide →](/docs/getting-started/how-it-works.md)
  
  
    Full reference for the Execute, Files, and MCP endpoints. Request/response schemas, error codes, and examples.

    [View API reference →](/docs/api/overview.md)

## All documentation pages (markdown)

- [MCP Protocol](https://baponi.ai/docs/api/mcp.md): Baponi MCP server reference. Model Context Protocol 2025-11-25 implementation. Streamable HTTP transport, sandbox_execute and files_download tools, stateless design.
- [Execute Code](https://baponi.ai/docs/api/execute.md): POST /v1/sandbox/execute. Run Python, Node.js, or Bash code in an isolated sandbox with sub-20ms overhead. Full request/response reference with examples.
- [Executions](https://baponi.ai/docs/api/executions.md): GET /v1/executions/{trace_id} and /output. Retrieve execution status, results, and reconnect to streaming output after a disconnect. Full request/response reference.
- [Files API](https://baponi.ai/docs/api/files.md): List, upload, download, and delete files in Baponi threads and volumes. Signed URLs for direct cloud storage transfers up to 10 GiB.
- [API Overview](https://baponi.ai/docs/api/overview.md): Baponi API reference: authentication with API keys, base URLs, Stripe-style error responses, rate limiting, and available SDKs. Everything you need to integrate sandboxed code execution into your AI agent.
- [LangChain Deep Agents](https://baponi.ai/docs/guides/deep-agents.md): Add sandboxed code execution to your LangChain Deep Agent in 3 lines. BaponiSandbox gives your agent file read/write, code execution, grep, and glob with persistent state, per-call overrides, real-time streaming, and lifecycle management.
- [How Baponi Works](https://baponi.ai/docs/getting-started/how-it-works.md): Baponi is a sandboxed code execution platform for AI agents. One API call, sub-20ms overhead, no lifecycle management. Understand the execution model, thread_id persistence, BYOB storage, connectors, and LLM integration.
- [Quickstart: First Execution in 60 Seconds](https://baponi.ai/docs/getting-started/quickstart.md): Get from zero to running code in a Baponi sandbox in under 60 seconds. Sign up, create an API key, and execute your first Python, Node.js, or Bash command via curl, Python SDK, or MCP.
- [Remote Claude Code Sessions](https://baponi.ai/docs/guides/claude-code.md): Run Claude Code in a sandboxed Baponi environment with automatic credential injection for Anthropic API keys and GitHub access. Clone repos, run headless agent sessions, and push results, all fully isolated.
- [Storage Path Scoping: Multi-Tenant Data Isolation for AI Agents](https://baponi.ai/docs/guides/storage-path-scoping.md): Control exactly which subdirectories AI agents can access in your S3, GCS, or Azure buckets. Three levels of path constraints - connection, API key, and per-request - enforce least-privilege access with zero application code.
- [Workload Identity Federation](https://baponi.ai/docs/guides/workload-identity-federation.md): Connect your GCS or S3 buckets to Baponi without sharing secrets, using cross-project/cross-account trust.