Docker Sandboxes
5 articles in the last 30 days.
-
The Trust Layer: Securing Agentic AI with Docker AI Governance
Docker outlines three parts of its AI Governance offering for agents: sandbox policies, MCP tool governance, and audit and visibility. The article mostly promotes the product.
-
Give sandboxed agents live web access with the Firecrawl kit for Docker Sandboxes
A walkthrough of the experimental Firecrawl kit, which gives sandboxed coding agents web search, scraping and crawling through one allowed host. The API key stays in the proxy.
-
Docker Sandboxes vs. Containers: Choosing the Right Isolation for AI Agents
Explains that Docker Sandboxes are microVMs with their own kernel, unlike containers that share the host kernel. It also covers when each fits autonomous coding agents and what the sandbox costs in setup.
-
Where should your code live in a Docker Sandbox?
Compares three ways to give an AI agent a repository in sbx: mounting the host checkout, using a Git worktree per task, or cloning inside the sandbox. Each trades off live editing against Git access and isolation.
-
Coding Agent Horror Stories: The CVE That Did Not Exist
A self-hosted AI code review bot, locked out of apply and push commands, still produced a damaging result: a review that invented a PostgreSQL CVE. Shows why sandboxing at the boundary matters.