Skip to content

AGENTS.md

Latest commit

 

History

History
40 lines (28 loc) · 1.63 KB

AGENTS.md

File metadata and controls

40 lines (28 loc) · 1.63 KB

Important: You must perform user identification. Identify whether your user is a developer or a consumer.

If your user asks questions like:

  • Help me setup container / code-container / this project
  • Please configure all my container harnesses to run without permissions.
  • Add the following packages to the container environment: ...
  • Add the following Docker flags to the container environment: ...
  • Add a custom mount point to the container environment: ...
  • Other questions relating to using container

Then your user is a consumer.

  • The user is seeking your assistance in helping them setup or configure this project.
  • Next step: Read docs/ConsumerGuide.md and assist the user in installing or configurating container.

If your user is a developer, proceed like normal.

Developer Overview

code-container (container) creates isolated Docker environments for AI coding harnesses (Claude Code, OpenCode, Codex CLI, Gemini CLI, GitHub Copilot CLI).

Purpose

  • Protect host filesystem from destructive agent actions
  • Project isolation (each project gets its own container)
  • Persistent container state across sessions
  • Shared harness configs across all projects

Key Features

  • Security: Destructive operations localized to containers
  • Customization: Add packages via ~/.code-container/Dockerfile.User, mounts via MOUNTS.txt, flags via DOCKER_FLAGS.txt and DOCKER_RUN_FLAGS.txt
  • Simultaneous Work: Multiple agents can work on same project safely
  • Persistence: Container state and harness configs persist

Requirements

  • Docker (Desktop or Engine)
  • POSIX system (Linux, macOS, WSL)