Code execution sandboxes
Executing code you did not write — from an LLM, a user, or a pull request — is a sandboxing problem first and a runtime problem second. These posts cover code interpreter architecture, Python and notebook execution, timeouts and resource limits, and why the isolation boundary matters more than the language runtime.
Code interpreter sandboxes on PandaStack212 posts
What FedRAMP Asks of a Platform That Runs Untrusted Code
An ephemeral-by-design fleet is hostile to a control set that assumes a stable, enumerable inventory — and a snapshot memory image is an asset class nobody wrote a control for.
Answering a Security Questionnaire When You Run Customer Code
The spreadsheet was written for a CRM. You run code a customer's model wrote ninety seconds ago. About a dozen rows carry the entire review, and the most important one is not on the sheet at all.
Giving a MicroVM Access to a Customer's Private Network
The customer's database is in their VPC and your sandbox is not. The naive answer is to hand them your egress IPs and ask them to open a hole; the answer that survives a security review is a WireGuard peer per sandbox, minted after restore, revoked on teardown, and never, ever baked into a snapshot.
Vendor Lock-In in Code Execution Infrastructure
Interface lock-in is an adapter and a bad afternoon. Data lock-in is a project. Semantic lock-in — you built on a behaviour nobody else sells — has no exit at all, which is why nobody sells you a mitigation for it. Written by a vendor, so I owe you the same audit of my own product.
What "persistent" actually means in a sandbox
You wrote the file. You ran cat and saw it. Neither of those facts says the bytes are on a disk. Here is every layer a write passes through inside a microVM, which of them a crash erases, and why the honest answer for an ephemeral rootfs is not "fsync harder" but "get the artifact out".
A plain-English glossary of sandbox and microVM terms
This vocabulary is genuinely confusing, and a lot of it is vendors using one word for three things. Here is the glossary I wish someone had handed me: 48 terms, each with the distinction that actually matters and the specific way people get it wrong.
Running Customer Trading Strategies in Isolated microVMs
Your customers write strategies. Your infrastructure runs them, next to each other, holding credentials that place orders. This is the untrusted-code problem with a P&L attached.
How to add human approval to agent code execution
The gate that asks about everything gets clicked through by Thursday. The useful design asks about almost nothing, because isolation earns you the right not to ask.
How to run SWE-agent in a sandbox
SWE-agent hands a language model a shell over your repository. On a laptop, that shell also reaches your SSH keys. Here is the isolated version.
More posts in this topic
- How to sandbox code from Cline and Continue
- How many sandboxes should a multi-agent system have?
- What is a deep research agent?
- How to give a Haystack agent a code execution tool
- How to hand off work between AI agents
- How to run a RAG pipeline in an ephemeral sandbox
- How to Sandbox OpenHands Agent Execution in a microVM
- How to Sandbox Aider's Command Execution
- How to give Goose a code execution sandbox
- Running a SWE-bench Evaluation Harness on microVMs
- Tracing agent runs when the code runs in a sandbox
- Sandboxing IDE Agent Extensions
- How to Sandbox a browser-use Agent
- What Coding Agents Actually Need From a Sandbox
- The Best Open-Source Coding Agents in 2026 — and What Each One Needs From the Machine It Runs On
- The best StackBlitz alternatives in 2026
- How to give a Strands agent a code execution tool
- OSS License Compliance Scanning, One MicroVM Per Scan
- Building Container Images for Untrusted Repos Without Privileged Docker
- SOC 2 for a Platform That Runs Other People's Code
- Compiling User-Submitted LaTeX Without Handing Over a Shell
- Running Customer Git Hooks in Isolated microVMs
- Top 7 Celery Hosting Platforms in 2026
- What is tool calling for AI agents?
- The best sandbox APIs for .NET agents in 2026
- How to Debug a Hung Process in a Sandbox
- How to Give a LangGraph Agent a Code Execution Tool
- Running npm install in a MicroVM: Dependency Installation Is Arbitrary Code Execution
- PandaStack vs Coder
- How to Vet a Code Execution Vendor's Security
- The best RunPod alternatives in 2026
- The best Beam Cloud alternatives in 2026
- Vercel Sandbox vs Cloudflare's Sandbox SDK
- The best sandbox APIs for Rust AI agents in 2026
- How to deploy a Streamlit app without Docker
- How to run Puppeteer in a sandbox
- How to give an Agno agent a code execution tool
- How to Give a Semantic Kernel Agent a Code Execution Tool
- How to Give a DSPy Program a Code Execution Sandbox
- Best Judge0 Alternatives (2026): Code Execution APIs Compared
- Best Flask Hosting Platforms (2026)
- Best Gradio Hosting Platforms (2026)
- How to Give an AutoGen Agent a Code Execution Tool
- How to Give a smolagents Agent a Code Execution Sandbox
- How to Expose a Sandbox Port on a Public URL
- How to give a Google ADK agent a code execution tool
- The best Cloudflare Containers alternatives in 2026
- The best Morph Cloud alternatives in 2026
- How to give a Mastra agent a code execution tool
- The best CodeSandbox alternatives in 2026
- How to upload and download files from a sandbox
- How to run Playwright tests in a sandbox
- A Checkpoint Is a Program: Isolating Untrusted Model Weights
- Cloudflare Sandbox SDK vs E2B for Agent Code
- How to give a LlamaIndex agent a code execution tool
- How to give a Pydantic AI agent a code execution tool
- node:vm Is Not a Sandbox (And Neither Was vm2)
- Customer-Authored dbt Runs, One microVM Per Tenant
- How to give the Claude Agent SDK a code execution sandbox
- The Best Sandboxes for Claude Agents in 2026
- How to give a LangChain agent a code execution tool
- How to add code execution to the OpenAI Agents SDK
- How to give a CrewAI agent a code execution tool
- How to add a code interpreter to the Vercel AI SDK
- How to control sandbox lifetime: TTL, idle, and cleanup
- Runnable Code Playgrounds in Your Docs, Safely
- Best Jupyter notebook hosting platforms in 2026
- The best Gitpod alternatives in 2026
- How to stream command output from a sandbox
- How to persist data in a sandbox
- Per-Partner Isolation for B2B EDI and File Integrations
- Modal vs Vercel Sandbox for Running Untrusted Code
- The best Streamlit hosting platforms in 2026
- The best Runloop alternatives in 2026
- Running live technical interviews on microVMs
- Modal vs Daytona for AI Agent Code Execution
- Vercel Sandbox vs E2B for AI-Generated Code
- The best Python hosting platforms in 2026
- The best Django hosting platforms in 2026
- How to deploy a Django app without Docker
- How to deploy a Flask app without Docker
- Running AI Bookkeeping Agents in Per-Tenant MicroVMs
- Isolating PII Redaction Pipelines in MicroVMs
- Sandboxing SBOM and Vulnerability Scans of Untrusted Artifacts
- The Best FastAPI Hosting Platforms in 2026
- Sandboxing AI Resume-Screening Agents With MicroVMs
- Per-Tenant MicroVM Isolation for Thumbnail Generation
- MicroVM Isolation for AI Invoice OCR and Extraction
- Running dotnet restore and dotnet build on untrusted code in a microVM
- Sandboxing AI incident-response agents: isolating the runbook
- AI Contract Review Agents: Isolating Privileged Documents in MicroVMs
- AI agents that edit podcasts and video: sandboxing the render pipeline
- PandaStack vs Replit
- Running bundle install on untrusted Ruby code in a microVM
- Deploying FastAPI without writing a Dockerfile
- Running RL environments in microVMs: isolating rollout workers
- Sandboxing an AI Agent That Operates Your Kubernetes Clusters
- Detonating malware in Firecracker microVMs
- Sandboxing untrusted uploads: ImageMagick, ffmpeg, LibreOffice
- Firecracker vs Firejail: confining apps vs containing strangers
- The Best Code Execution Sandboxes for Education Platforms in 2026
- Firecracker vs WebContainers: where should untrusted code run?
- Letting an AI Agent Run Integration Tests Against Real APIs
- Running User-Generated Game Mods in Isolated microVMs
- Running IoT and Embedded Firmware Emulation in Disposable MicroVMs
- Isolating Per-Tenant Geospatial Processing Jobs in MicroVMs
- Sandboxing AI-Agent 3D Rendering and Asset Pipelines in MicroVMs
- Running Fuzzing Harnesses and Crash Reproduction in MicroVMs
- Isolating Per-Tenant CSV and Bulk Import Pipelines in MicroVMs
- PandaStack vs GitHub Codespaces
- Per-Tenant AI Voice Transcription in MicroVMs
- Running Per-Tenant Billing and Usage Metering in MicroVMs
- How to Sandbox an Untrusted composer install
- Best Sandbox APIs for Python Coding Agents in 2026
- PandaStack vs Koyeb: an honest comparison
- Sandboxing User-Written Webhook Transformations
- PandaStack vs AWS Lambda for Running Untrusted / LLM-Generated Code
- The Best AI Code Execution Platforms in 2026
- Running a WebAssembly Plugin Host Inside a MicroVM
- How Sandbox Pricing Models Actually Work in 2026
- Per-Tenant MicroVM Isolation for PDF and Invoice Generation
- Sandboxing AI-Agent Spreadsheet Automation
- Isolating Multi-Tenant GenAI Image Jobs in MicroVMs
- A Production Checklist for Running Untrusted Code Safely
- Give Your AI Agent a Terminal — Not Yours
- Run AI-Agent DB Migrations in an Isolated microVM
- Firecracker vs nsjail: which for running untrusted code?
- Firecracker vs Bubblewrap: sandboxing untrusted code
- Per-Tenant ML Inference Isolation With MicroVMs
- Sandbox Your Agent's Computer-Vision Pipeline in MicroVMs
- Per-Tenant ETL Pipeline Isolation With MicroVMs
- Firecracker vs E2B: You're Comparing Two Layers
- Best Vercel Sandbox Alternatives in 2026
- PandaStack vs Freestyle: Two Bets on AI-Generated Code
- Run Code-Migration Agents in MicroVMs, Not on Your CI Box
- The Best Modal Alternatives in 2026
- Sandboxing an untrusted Java/Maven (and Gradle) build
- Firecracker vs the Nix build sandbox: different jobs
- How to Sandbox AI Agents in 2026
- Top 5 AI Agent Sandbox Platforms in 2026
- Per-Tenant LLM Fine-Tuning Jobs in Isolated microVMs
- Sandboxing PDF & Document Processing for AI Agents
- Per-Tenant Scheduled Report Generation in microVMs
- Best gVisor Alternatives in 2026
- Building untrusted Rust code in a microVM
- A Zero-Trust Architecture for Executing Untrusted Code
- Per-Tenant Object Storage Isolation with microVMs
- Long-Running Sandboxes for AI Agents
- Stateful vs Ephemeral AI Agent Sandboxes
- Resuming & Reconnecting AI Agent Sessions
- What Is an AI Agent Sandbox?
- How AI Agent Sandboxes Work
- Why Every AI Agent Needs a Sandbox
- Red-Teaming LLMs in Disposable microVMs
- Building an autograder that runs student code safely in microVMs
- A microVM Harness for SWE-bench & terminal-bench
- Building untrusted Go code in a microVM
- Per-Tenant Analytics Queries in Isolated microVMs
- Isolating AI Shopping Agents in MicroVMs
- Running npm install on untrusted code in a microVM
- Per-Match Game Server Isolation on MicroVMs
- Best Secure Code Execution APIs in 2026
- Sandboxing AI-Generated Terraform and Infrastructure-as-Code
- Running AI Agent Eval Harnesses in Isolated microVMs
- Testing LLM-Generated Database Migrations Safely in a Sandbox
- Building an Online Judge on microVMs
- Multi-Tenant Notebooks: microVM Sandboxes vs JupyterHub
- Sandboxing an AI Email-Triage Agent
- Run AI-Agent-Built Data Pipelines in Isolated microVMs
- Running LLM-Generated SQL and Analysis Safely
- Run AI Browser Agents in Isolated MicroVMs
- The Best Firecracker Sandbox APIs in 2026
- PandaStack vs E2B for Building a Code Interpreter
- Running AI Pentest Agents in Disposable MicroVMs
- PandaStack vs Fly.io Machines: an honest comparison
- The Best E2B Alternatives in 2026
- How to Run AI-Generated Code Safely: a 2026 Playbook
- Per-User Notebook Kernels: Isolating a Hosted Jupyter Product
- Give Your AI Coding Agent a Sandbox to Run PR Checks
- WebAssembly vs Firecracker for Untrusted Code
- Sandboxing LLM Batch Post-Processing at Scale
- Safely running pip install from LLM-generated code
- A Sandbox for AI Agent Computer Use
- Best Self-Hosted Code Execution Sandboxes in 2026
- Controlling Network Egress for Untrusted Code
- How to Sandbox Untrusted Jupyter Notebooks Per User
- Sandboxed Web Scraping for AI Agents
- PandaStack vs Cloudflare Workers for Untrusted Code
- How to Jail LLM-Generated Code
- Build an AI Data Analyst That Runs Code on User Data
- Safely Running Shell Commands an AI Agent Decides to Execute
- How to Build a Sandboxed AI Coding Agent (2026)
- Best Code Execution Sandboxes for AI Agents (2026)
- Code Interpreter API Pricing, Compared
- How to Sandbox Untrusted & AI-Generated Code
- Why Docker Isn't a Sandbox
- Multi-Tenant Code Execution: Isolation Requirements
- An Open-Source OpenAI Code Interpreter Alternative
- How to run untrusted (and AI-generated) code safely
- Secure code execution for AI agents: isolation, ephemerality, and network control
- How to Give Your AI Agent a Sandbox (With Code)
- PandaStack vs Daytona for AI Sandboxes
- Build a Code Interpreter with a Python Sandbox