agnes-the-ai-analyst/docs/setup/bootstrap.yaml

version: "1.0"
project_name: "ai_data_analyst"
project_dir: "."

# Placeholders filled by webapp per-user:
#   {server_host}  - server IP or hostname
#   {ssh_alias}    - SSH config alias (default: "data-analyst", configurable to avoid conflicts)
#   {ssh_key}      - SSH private key path (default: ~/.ssh/data_analyst_server)
#   {username}     - analyst username on server
#   {webapp_url}   - webapp URL for registration

server:
  host: "{server_host}"
  hostname: "{ssh_alias}"
  webapp_url: "{webapp_url}"

setup:
  steps:
    - name: "detect_existing_project"
      description: "Check if project already exists"
      check: "test -f ./CLAUDE.md"
      on_success: "verify_project_identity"
      message: |
        Existing CLAUDE.md detected in current directory.
        Verifying this is an AI Data Analyst project...        

    - name: "verify_project_identity"
      description: "Verify this is the correct project type"
      check: "grep -q 'AI Data Analyst' ./CLAUDE.md"
      on_success: "existing_project_confirmed"
      on_failure: |
        Wrong project type detected.
        The CLAUDE.md file exists but doesn't match AI Data Analyst.

        Options:
        - Choose a different directory for setup
        - Remove existing CLAUDE.md if this was a mistake        
      message: |
        AI Data Analyst project confirmed.

        This directory is already set up. You can:
        - Sync latest data: bash server/scripts/sync_data.sh
        - View project context: cat CLAUDE.md

        To recreate CLAUDE.md: rm -f ./CLAUDE.md and re-run bootstrap        

    - name: "check_directory_empty"
      description: "Warn if directory is not empty"
      check: "[ $(ls -A . 2>/dev/null | wc -l) -eq 0 ]"
      on_failure: |
        Current directory is not empty.

        This setup will create:
        - .claude/ (project metadata)
        - server/ (read-only data from server: parquet files, docs, scripts)
        - user/ (your workspace: DuckDB database, artifacts)
        - .venv/ (Python virtual environment)

        Make sure you're in the correct directory before continuing.        
      warn_only: true
      message: |
        Starting setup in current directory...        

    - name: "generate_ssh_key"
      description: "Generate SSH key for server authentication"
      check: "test -f {ssh_key}.pub"
      action: |
        ssh-keygen -t ed25519 -f {ssh_key} -C "{username}@{ssh_alias}" -N ''        
      on_success: "show_public_key"
      message: |
        SSH key generated successfully.        

    - name: "show_public_key"
      description: "Display SSH public key to user"
      action: "cat {ssh_key}.pub"
      message: |
        Your SSH public key has been generated!

        Next steps:
        1. Copy the public key shown above
        2. Go to: {webapp_url}
        3. Sign in and paste the key into the form
        4. Wait a few seconds for account creation
        5. Come back here to continue        
      wait_for_user: true

    - name: "add_ssh_config"
      description: "Add SSH config entry"
      requires: ["show_public_key"]
      check: "ssh -o ConnectTimeout=5 -o BatchMode=yes {ssh_alias} echo ok 2>/dev/null"
      action: |
        mkdir -p ~/.ssh

        # Check if alias already exists with a different host
        if grep -q 'Host {ssh_alias}' ~/.ssh/config 2>/dev/null; then
          EXISTING_HOST=$(awk '/Host {ssh_alias}/,/Host / {if (/HostName/) print $2}' ~/.ssh/config | head -1)
          if [[ "$EXISTING_HOST" != "{server_host}" ]]; then
            echo "WARNING: SSH alias '{ssh_alias}' already exists pointing to $EXISTING_HOST"
            echo "Skipping SSH config - please resolve manually or use a different alias."
            exit 1
          fi
        else
          cat >> ~/.ssh/config << 'EOF'

        Host {ssh_alias}
            HostName {server_host}
            User {username}
            IdentityFile {ssh_key}
            StrictHostKeyChecking accept-new
        EOF
          chmod 600 ~/.ssh/config
        fi        
      message: |
        SSH configuration added for {ssh_alias} server.        

    - name: "test_ssh_connection"
      description: "Test SSH connection to server"
      requires: ["add_ssh_config"]
      action: "ssh -o ConnectTimeout=5 -o BatchMode=yes {ssh_alias} echo 'ok' 2>/dev/null"
      message: |
        Testing connection to data server...
        Connection successful!        
      on_failure: |
        SSH connection failed!

        Please verify:
        1. You completed registration at {webapp_url}
        2. Your account was created successfully
        3. Your username matches: {username}        
      retry: true
      max_retries: 3

    - name: "create_folders"
      description: "Create local project structure"
      action: |
        mkdir -p ./server/docs ./server/scripts ./server/examples ./server/parquet ./server/metadata
        mkdir -p ./user/duckdb ./user/notifications ./user/artifacts ./user/scripts ./user/parquet ./user/sessions        
      message: |
        Project structure created (server/, user/).        

    - name: "check_rsync"
      description: "Verify rsync is available"
      check: "command -v rsync >/dev/null 2>&1"
      warn_only: true
      on_failure: |
        rsync is not installed. Install it for better sync performance:

        macOS:   brew install rsync
        Ubuntu:  sudo apt-get install -y rsync
        RHEL:    sudo yum install -y rsync

        Without rsync, scp will be used as fallback (slower).        

    - name: "download_server_data"
      description: "Download all server data (scripts, docs, metadata, parquet)"
      action: |
        echo "Syncing scripts..."
        rsync -avz --no-perms --no-group {ssh_alias}:server/scripts/ ./server/scripts/ 2>/dev/null || \
          scp -r {ssh_alias}:server/scripts/* ./server/scripts/ 2>/dev/null || true

        echo "Syncing documentation..."
        rsync -avz --no-perms --no-group {ssh_alias}:server/docs/ ./server/docs/ 2>/dev/null || \
          scp -r {ssh_alias}:server/docs/* ./server/docs/ 2>/dev/null || true

        echo "Syncing examples..."
        rsync -avz --no-perms --no-group {ssh_alias}:server/examples/ ./server/examples/ 2>/dev/null || true

        echo "Syncing metadata..."
        rsync -avz --no-perms --no-group {ssh_alias}:server/metadata/ ./server/metadata/ 2>/dev/null || \
          scp -r {ssh_alias}:server/metadata/* ./server/metadata/ 2>/dev/null || true

        echo "Syncing parquet data (this may take a few minutes)..."
        rsync -avz --no-perms --no-group --progress {ssh_alias}:server/parquet/ ./server/parquet/ 2>/dev/null || \
          scp -r {ssh_alias}:server/parquet/* ./server/parquet/ 2>/dev/null || true        
      requires: ["test_ssh_connection", "create_folders"]
      message: |
        Downloading data from server...
        Data downloaded successfully!        

    - name: "setup_venv"
      description: "Create Python virtual environment and install dependencies"
      check: "test -f ./.venv/bin/python || test -f ./.venv/Scripts/python.exe"
      action: |
        if command -v python3 >/dev/null 2>&1; then
          PYTHON_CMD=python3
        else
          PYTHON_CMD=python
        fi

        $PYTHON_CMD -m venv ./.venv

        if [ -f ./.venv/bin/activate ]; then
          source ./.venv/bin/activate
        else
          source ./.venv/Scripts/activate
        fi

        pip install --upgrade pip --quiet
        pip install pandas pyarrow duckdb pyyaml python-dotenv --quiet        
      requires: ["create_folders"]
      message: |
        Setting up Python environment...
        Python environment ready!        

    - name: "initialize_duckdb"
      description: "Initialize DuckDB views on Parquet files"
      action: |
        if [[ -f server/scripts/setup_views.sh ]]; then
          bash server/scripts/setup_views.sh
        else
          echo "setup_views.sh not found, skipping DuckDB initialization"
        fi        
      requires: ["download_server_data", "setup_venv"]
      message: |
        Initializing DuckDB analytical database...
        DuckDB initialized! All tables ready for queries.        

    - name: "setup_claude_project_context"
      description: "Create Claude Code project context files"
      action: |
        # Generate CLAUDE.md from template
        if [[ -f "./server/docs/setup/claude_md_template.txt" ]]; then
          sed -e "s/{username}/{username}/g" \
              ./server/docs/setup/claude_md_template.txt > ./CLAUDE.md
          chmod 644 ./CLAUDE.md
        fi

        # Create CLAUDE.local.md for personal customizations
        if [[ ! -f "./CLAUDE.local.md" ]]; then
          cat > ./CLAUDE.local.md << 'LOCALEOF'
        # CLAUDE.local.md

        Your personal instructions for Claude Code in this project.
        This file is NOT overwritten by data sync - it is yours to customize.

        ## Your Custom Instructions

        Add your preferences, shortcuts, or project-specific notes below:

        LOCALEOF
          chmod 644 ./CLAUDE.local.md
        fi

        # Copy project permissions
        mkdir -p ./.claude
        if [[ -f "./server/docs/setup/claude_settings.json" ]]; then
          cp ./server/docs/setup/claude_settings.json ./.claude/settings.json
        fi        
      requires: ["download_server_data"]
      message: |
        CLAUDE.md created (auto-updated on sync).
        CLAUDE.local.md created (your personal customizations, never overwritten).        

    - name: "check_setup"
      description: "Verify setup completed successfully"
      requires: ["initialize_duckdb", "setup_claude_project_context"]
      message: |
        Setup complete! Your AI Data Analyst environment is ready.

        What's been set up:
        - Data tables synced as local Parquet files
        - DuckDB analytical database with views configured
        - Python environment with pandas, pyarrow, duckdb
        - Helper scripts for data sync

        You can now start asking questions about your data.
        See server/docs/data_description.md for table schemas.

        To sync latest data: bash server/scripts/sync_data.sh        

# Python dependencies
dependencies:
  - pandas>=2.0.0
  - pyarrow>=12.0.0
  - duckdb>=0.9.0
  - pyyaml>=6.0
  - python-dotenv>=1.0.0