Refocus container as Linux-only GAIA runtime.

Remove AI IDE tooling references and align docs/tests with PyPI install flow.

Author: itomek

.cursor/rules/docker.mdc

@@ -1,5 +1,5 @@
 ---
-description: Docker and container-specific development rules
+description: Docker and container-specific rules
 globs:
   - Dockerfile
   - docker-compose.yml
@@ -16,8 +16,8 @@ globs:
 
 ## Image Architecture
 - Base: Python 3.12 slim
-- Includes: Node.js, zsh, git, Claude Code CLI
-- Entrypoint clones GAIA and installs dependencies at runtime
+- Includes: Node.js, zsh, git
+- Entrypoint installs GAIA from PyPI at runtime
 - Uses uv for 10-100x faster Python package installation
 
 ## Multi-Instance Support
@@ -27,11 +27,9 @@ The container is designed to spawn multiple instances:
 - Different container names for parallel development
 
 ## Volume Management
-- `/home/gaia/.claude`: Persists Claude Code credentials
-- `/source/gaia`: GAIA repository (cloned at startup, not bind-mounted)
+- `/source`: Working directory
 - `/host`: Optional host directory mount for file exchange
 
 ## Performance
-- First install: ~2-3 minutes (including git clone)
+- First install: ~2-3 minutes
 - Reinstall with cache: ~30 seconds
-- YOLO reset: ~2-3 minutes (fresh clone + install)

.cursor/rules/index.mdc

@@ -1,12 +1,12 @@
 ---
-description: Main project rules for GAIA Docker development
+description: Main project rules for GAIA Docker
 alwaysApply: true
 ---
 
 # GAIA Docker Project Rules
 
 ## Project Overview
-This is a Docker containerization project for AMD GAIA (AI framework) development. The project creates an isolated, disposable development environment optimized for "YOLO mode" - where containers can be freely destroyed and recreated.
+This is a Linux Docker container for AMD GAIA (AI framework). GAIA is installed from PyPI at container startup.
 
 ## Tech Stack
 - **Base**: Docker & Docker Compose
@@ -16,26 +16,24 @@ This is a Docker containerization project for AMD GAIA (AI framework) developmen
 - **Target Application**: AMD GAIA framework
 
 ## Key Files
-- `Dockerfile`: Container image definition with Python 3.12, Node.js, and dev tools
+- `Dockerfile`: Container image definition with Python 3.12 and tools
 - `docker-compose.yml`: Service configuration with environment variables
-- `entrypoint.sh`: Container startup script (clones GAIA, installs dependencies)
+- `entrypoint.sh`: Container startup script (installs GAIA from PyPI)
 - `pyproject.toml`: Python project configuration with test dependencies
 - `tests/`: Integration and unit tests using pytest
 
 ## Container Architecture
-- GAIA is cloned INSIDE the container (not bind-mounted) for complete isolation
-- Credentials persist in Docker volumes (`/home/gaia/.claude`)
-- Container is designed to be disposable - fast recreation (~2-3 minutes)
+- GAIA is installed from PyPI at startup
+- Container is isolated and reproducible
 - Uses uv for Python package installation (much faster than pip)
 
 ## Environment Variables
 Key env vars in docker-compose.yml:
-- `GAIA_BRANCH`: Branch to clone (default: main)
-- `SKIP_INSTALL`: Skip pip install on restart (for faster iteration)
+- `LEMONADE_URL`: Lemonade server URL (default: http://localhost:5000/api/v1)
+- `GAIA_VERSION`: GAIA version to install from PyPI
+- `SKIP_INSTALL`: Skip pip install on restart (faster startup)
 - Ports: LEMONADE_PORT (5000), GAIA_API_PORT (8000), WEB_PORT (3000)
 
 ## Important Notes
-- This is a development container - optimized for speed, not production security
-- Credentials (Claude Code OAuth) persist in volumes, not in container
-- The container is "YOLO safe" - designed to be destroyed and recreated freely
+- This is a Linux container for GAIA
 - When suggesting changes to Dockerfile or entrypoint.sh, ensure tests still pass

.cursor/rules/python.mdc

@@ -41,6 +41,5 @@ uv run pytest
 ## Inside Container
 The container uses `uv pip install` for installing GAIA:
 ```bash
-cd /source/gaia
-uv pip install -e ".[dev,mcp,eval,rag]"
+uv pip install --system "amd-gaia[dev,mcp,eval,rag]==<version>"
 ```

.cursorignore

@@ -37,9 +37,6 @@ Thumbs.db
 # Logs
 *.log
 
-# Docker volumes (if mounted locally)
-.claude/
-
 # Build artifacts
 dist/
 build/

.env.example

@@ -68,5 +68,5 @@ jobs:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
           repository: ${{ env.IMAGE_NAME }}
-          short-description: "Isolated development container for AMD GAIA framework"
+          short-description: "GAIA Linux container"
           readme-filepath: ./docs/docker-hub-description.md

.github/workflows/publish.yml

@@ -17,7 +17,6 @@ htmlcov/
 
 # Docker
 .docker/
-*.env
 
 # IDEs
 .vscode/

.gitignore

@@ -1,5 +1,5 @@
-# GAIA Development Container
-# Provides isolated Python 3.12 environment with all tools for GAIA development
+# GAIA Linux Container
+# Provides isolated Python 3.12 environment for GAIA
 # GAIA is installed from PyPI at runtime based on version
 
 FROM python:3.12-slim
@@ -52,13 +52,6 @@ RUN useradd -m -s /bin/zsh $USERNAME && \
 RUN mkdir -p /source /host && \
     chown -R $USERNAME:$USERNAME /source /host
 
-# Install git-delta (better git diffs)
-ARG GIT_DELTA_VERSION=0.18.2
-RUN ARCH=$(dpkg --print-architecture) && \
-    wget -q "https://github.com/dandavison/delta/releases/download/${GIT_DELTA_VERSION}/git-delta_${GIT_DELTA_VERSION}_${ARCH}.deb" && \
-    dpkg -i "git-delta_${GIT_DELTA_VERSION}_${ARCH}.deb" && \
-    rm "git-delta_${GIT_DELTA_VERSION}_${ARCH}.deb"
-
 # Copy entrypoint script
 COPY entrypoint.sh /usr/local/bin/entrypoint.sh
 RUN chmod +x /usr/local/bin/entrypoint.sh
@@ -75,9 +68,6 @@ RUN sh -c "$(wget -O- https://github.com/deluan/zsh-in-docker/releases/download/
 # Install uv (fast Python package installer)
 RUN curl -LsSf https://astral.sh/uv/install.sh | sh
 
-# Install Claude Code CLI (interactive OAuth login)
-RUN curl -fsSL https://claude.ai/install.sh | bash
-
 # Configure environment
 ENV SHELL=/bin/zsh
 ENV PATH="/home/gaia/.cargo/bin:/home/gaia/.local/bin:$PATH"