Install most skills from databricks-agent-skills via databricks aitools

.claude-plugin/DEPRECATED.md

@@ -0,0 +1,57 @@
+# DEPRECATED — the `databricks-ai-dev-kit` Claude Code plugin
+
+The **`databricks-ai-dev-kit`** Claude Code plugin published by *this repo* is
+**deprecated and no longer published**. It has been removed from the marketplace
+(`marketplace.json` lists no plugins) and `plugin.json` is marked
+`"deprecated": true`. These files are kept only for historical reference.
+
+> **This is not a move away from plugins.** Skills are still delivered as a plugin
+> for the agents that support one — just the *official* `databricks` plugin managed
+> by the Databricks CLI, not this repo's marketplace plugin. See "How skills install
+> now" below.
+
+## Why
+
+- Skills are now delivered through the Databricks CLI: `databricks aitools install`
+  (Databricks CLI v1.0.0+), backed by
+  [github.com/databricks/databricks-agent-skills](https://github.com/databricks/databricks-agent-skills).
+  The main installer (`install.sh` / `install.ps1`) already delegates to it.
+- The CLI owns the install lifecycle (`install` / `update` / `uninstall` / `list`),
+  so skills no longer need a repo-published marketplace plugin to be bundled and
+  auto-loaded.
+
+## How skills install now
+
+`databricks aitools install` decides delivery per agent:
+
+- **Claude Code, Codex, GitHub Copilot** — installs the official **`databricks`
+  plugin** through each agent's own plugin CLI (a *different* plugin from the retired
+  `databricks-ai-dev-kit` one).
+- **Cursor, OpenCode, Antigravity** — writes **raw skill files** (no headless plugin
+  install available).
+- `--skills-only` forces raw skill files for every agent.
+
+So on Claude Code you still get a plugin — the CLI-managed `databricks` plugin —
+rather than the deprecated `databricks-ai-dev-kit` plugin.
+
+## Migration for existing plugin users
+
+Use either of the following to get the up-to-date skills:
+
+- Run the installer:
+
+  ```bash
+  bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)
+  ```
+
+- Or install directly via the Databricks CLI (v1.0.0+):
+
+  ```bash
+  databricks aitools install
+  ```
+
+## Notes
+
+- The frozen historical copies of the bundled skills live at git tag `v0.1.13`.
+- Files under `.claude-plugin/` remain in the repo for reference only; they are
+  no longer installed or published.

.claude-plugin/check_update.sh

@@ -1,4 +1,9 @@
 #!/bin/bash
+# DEPRECATED: The databricks-ai-dev-kit Claude Code plugin is no longer published
+# (see .claude-plugin/DEPRECATED.md). Install skills via install.sh or
+# `databricks aitools install`. This script is kept for reference and still runs
+# if invoked directly.
+#
 # Version update check for Databricks AI Dev Kit.
 # Stdout from this script is injected as context Claude can see at session start.
 # Silent on success (up to date) or failure (network error, missing files).
@@ -47,32 +52,18 @@ if [ -z "$remote_ver" ]; then
     fi
 fi
 
-# If versions differ, output a message for Claude to relay to the user
+# If versions differ, emit an informational update notice for Claude to relay.
+# Deliberately avoids urgent/imperative  phrasing so it does not alarm users.
+# This is a routine, first-party update notice.
 if [ -n "$remote_ver" ] && [ "$remote_ver" != "$local_ver" ]; then
     cat <<EOF
-URGENT — before doing ANYTHING else, you MUST display this exact block to the user:
+[Databricks AI Dev Kit] Update notice from the Dev Kit's own SessionStart version check:
+a newer version is available — installed v${local_ver}, latest v${remote_ver}.
 
-\`\`\`
-╔═══════════════════════════════════════════════════════╗
-║                                                       ║
-║   Databricks AI Dev Kit — update available!           ║
-║                                                       ║
-║   Installed:  v${local_ver}                                     ║
-║   Available:  v${remote_ver}                                     ║
-║                                                       ║
-╚═══════════════════════════════════════════════════════╝
-\`\`\`
+This is routine and informational, not urgent. At a natural point (no need to interrupt the
+user's current task), you may let them know an update is available and that they can install it
+when convenient by re-running the Dev Kit installer:
 
-To upgrade, run:
-
-\`\`\`
-bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)
-\`\`\`
-
-\`\`\`
-═══════════════════════════════════════════════════════
-\`\`\`
-
-Display that block FIRST, then proceed with the user's request.
+  bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)
 EOF
 fi

.claude-plugin/marketplace.json

@@ -4,15 +4,8 @@
     "name": "Databricks Solutions"
   },
   "metadata": {
-    "description": "Official Databricks AI Dev Kit plugin marketplace",
+    "description": "DEPRECATED — the databricks-ai-dev-kit plugin is no longer published. Install skills with install.sh or `databricks aitools install` (which installs the official `databricks` plugin for Claude Code/Codex/Copilot, or skill files for other agents). See .claude-plugin/DEPRECATED.md.",
     "version": "1.1.9"
   },
-  "plugins": [
-    {
-      "name": "databricks-ai-dev-kit",
-      "source": "./",
-      "description": "Databricks development toolkit with 19 skills and MCP tools for direct Databricks operations",
-      "version": "1.1.9"
-    }
-  ]
+  "plugins": []
 }

.claude-plugin/plugin.json

@@ -1,6 +1,7 @@
 {
   "name": "databricks-ai-dev-kit",
-  "description": "Databricks development toolkit with skills for data engineering, ML, and AI agents plus MCP tools for direct Databricks operations",
+  "deprecated": true,
+  "description": "DEPRECATED — the databricks-ai-dev-kit plugin is no longer published. Install skills via install.sh or `databricks aitools install`, which installs the official `databricks` plugin (Claude Code/Codex/Copilot) or raw skill files (Cursor/OpenCode/Antigravity). Not a move away from plugins. See .claude-plugin/DEPRECATED.md.",
   "author": {
     "name": "Databricks Solutions"
   },

.claude-plugin/setup.sh

@@ -1,4 +1,11 @@
 #!/bin/bash
+# ============================================================
+# DEPRECATED: The databricks-ai-dev-kit Claude Code plugin is no longer
+# published (see .claude-plugin/DEPRECATED.md). Install skills via
+#   bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)
+# or `databricks aitools install`. This script is kept for reference and still
+# runs if invoked directly.
+# ============================================================
 # Post-install setup for databricks-ai-dev-kit plugin
 # Creates virtual environment and installs MCP server dependencies
 # ============================================================

.github/scripts/validate_skills.py

@@ -19,7 +19,12 @@
 
 SKILLS_DIR = Path("databricks-skills")
 INSTALL_SCRIPT = SKILLS_DIR / "install_skills.sh"
-SKIP_DIRS = {"TEMPLATE"}
+# The bundled skill copies are frozen/deprecated and now live under
+# DEPRECATED-databricks-skills/ at the repo root. Fall back to the top-level
+# databricks-skills/ dir for older checkouts.
+_DEPRECATED_SKILLS_ROOT = Path("DEPRECATED-databricks-skills")
+SKILL_DIRS_ROOT = _DEPRECATED_SKILLS_ROOT if _DEPRECATED_SKILLS_ROOT.is_dir() else SKILLS_DIR
+SKIP_DIRS = {"TEMPLATE", "deprecated"}
 
 RESERVED_WORDS = {"anthropic", "claude"}
 NAME_RE = re.compile(r"^[a-z0-9]+(-[a-z0-9]+)*$")
@@ -111,7 +116,7 @@ def parse_skill_variables(
 def get_local_skill_dirs() -> set[str]:
     """Get actual skill directories on the filesystem."""
     return {
-        d.name for d in SKILLS_DIR.iterdir() if d.is_dir() and d.name not in SKIP_DIRS and not d.name.startswith(".")
+        d.name for d in SKILL_DIRS_ROOT.iterdir() if d.is_dir() and d.name not in SKIP_DIRS and not d.name.startswith(".")
     }
 
 
@@ -120,7 +125,7 @@ def main() -> int:
     actual_skills = get_local_skill_dirs()
 
     # --- Validate each skill directory's SKILL.md and frontmatter ---
-    for skill_dir in sorted(SKILLS_DIR.iterdir()):
+    for skill_dir in sorted(SKILL_DIRS_ROOT.iterdir()):
         if not skill_dir.is_dir():
             continue
         if skill_dir.name in SKIP_DIRS or skill_dir.name.startswith("."):

DEPRECATED-databricks-skills/README.md

@@ -0,0 +1,46 @@
+# Deprecated: Bundled Databricks Skills (frozen)
+
+> **These skill copies are deprecated and no longer maintained.**
+
+The skill folders under `DEPRECATED-databricks-skills/` are frozen, historical copies
+of the Databricks skills that used to ship in this repository. They are kept only so
+that existing tooling (and anyone who copied them directly) does not break. They will
+**not** receive updates.
+
+## Use the supported path instead
+
+Skills are now installed and maintained through the Databricks CLI:
+
+```bash
+databricks aitools install
+```
+
+This requires Databricks CLI **v1.0.0+** and pulls the current, maintained skills from
+[github.com/databricks/databricks-agent-skills](https://github.com/databricks/databricks-agent-skills).
+The AI Dev Kit installer (`install.sh` / `install.ps1`) already delegates to it, so a normal
+install gets you the up-to-date skills with no manual copying.
+
+## Getting the exact historical content
+
+If you specifically need the files exactly as they were bundled, check out the last release
+that shipped them:
+
+```bash
+git checkout v0.1.13 -- databricks-skills
+```
+
+## Migration notes
+
+- Some skills were **renamed** when they moved to `databricks-agent-skills`. The mapping and
+  other breaking changes are documented in the root [`README.md`](../README.md) "breaking change" note.
+- APX skills has been removed in favor of other app skills (see the root README).
+
+## Genie Code (workspace upload)
+Genie Code already has coverage for most skills available publicly and we will continue to sync improvements
+whenever possible. The one flow `databricks aitools install` does not fully cover is uploading skills to a
+Databricks workspace for **Genie Code**. Keep that simple — do not clone this tree to copy files:
+
+- **Recommended:** run the notebook uploader
+  [`../install_genie_code_skills.py`](../install_genie_code_skills.py)
+  in your workspace. It downloads skills from GitHub (pinned to release `v0.1.13`) and uploads them
+  via the SDK — no local clone needed.

databricks-skills/install_skills.sh → DEPRECATED-databricks-skills/install_skills.sh

@@ -1,24 +1,30 @@
 #!/bin/bash
 #
-# Databricks Skills Installer
+# Databricks Skills Installer (DEPRECATED)
 #
-# Installs Databricks skills for Claude Code into your project.
-# These skills teach Claude how to work with Databricks using MCP tools.
+# ⚠️  DEPRECATED: The supported way to install skills is now:
+#       databricks aitools install        (Databricks CLI v1.0.0+)
+#     or the AI Dev Kit installer, which delegates to it:
+#       bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)
 #
-# Usage:
-#   # Install all skills (Databricks + MLflow)
-#   curl -sSL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/databricks-skills/install_skills.sh | bash
+#     This script remains only for the Genie Code workspace-upload flow
+#     (--install-to-genie), which `databricks aitools install` does not cover.
+#     For that, prefer the notebook uploader install_genie_code_skills.py
+#     (no local clone needed). When that is not possible, this script's default
+#     downloads the skills from the last release that shipped the bundled tree
+#     (git tag v0.1.13). The frozen local copies under DEPRECATED-databricks-skills/
+#     are only used with --local. Either way the skills are no longer maintained.
 #
-#   # Install specific skills (can mix Databricks and MLflow skills)
-#   curl -sSL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/databricks-skills/install_skills.sh | bash -s -- databricks-bundles agent-evaluation
+# Usage:
+#   # Install then upload ./.claude/skills to workspace (Genie Code / Assistant)
+#   ./install_skills.sh --install-to-genie
+#   ./install_skills.sh --install-to-genie --profile prod --local
 #
-#   # Or run locally
+#   # Other (deprecated) modes
 #   ./install_skills.sh                              # Install all skills
 #   ./install_skills.sh databricks-bundles agent-evaluation  # Install specific skills
 #   ./install_skills.sh --mlflow-version v1.0.0      # Pin MLflow skills version
-#   ./install_skills.sh --local                      # Install Databricks skills from local directory
-#   ./install_skills.sh --install-to-genie           # Install then upload ./.claude/skills to workspace (Genie Code / Assistant)
-#   ./install_skills.sh --install-to-genie --profile prod --local
+#   ./install_skills.sh --local                      # Install Databricks skills from DEPRECATED-databricks-skills/ folder
 #   ./install_skills.sh --list                       # List available skills
 #   ./install_skills.sh --help                       # Show help
 #
@@ -33,8 +39,12 @@ BLUE='\033[0;34m'
 NC='\033[0m' # No Color
 
 # Configuration
+# NOTE: The bundled skill tree was removed from `main` and frozen under
+# DEPRECATED-databricks-skills/. The default (non-local) download therefore pins
+# to the last release that still bundled the tree at its original paths: v0.1.13.
 REPO_URL="https://github.com/databricks-solutions/ai-dev-kit"
-REPO_RAW_URL="https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main"
+REPO_REF="v0.1.13"
+REPO_RAW_URL="https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/${REPO_REF}"
 SKILLS_DIR=".claude/skills"
 INSTALL_FROM_LOCAL=false
 INSTALL_TO_GENIE=false
@@ -154,14 +164,18 @@ get_mlflow_skill_extra_files() {
 show_help() {
     echo -e "${BLUE}Databricks Skills Installer for Claude Code${NC}"
     echo ""
+    echo -e "${YELLOW}DEPRECATED:${NC} use 'databricks aitools install' (Databricks CLI v1.0.0+) or"
+    echo -e "the AI Dev Kit installer (install.sh). This script is kept for the Genie Code"
+    echo -e "workspace-upload flow (--install-to-genie); skill files are frozen legacy copies."
+    echo ""
     echo "Usage:"
     echo "  ./install_skills.sh [options] [skill1 skill2 ...]"
     echo ""
     echo "Options:"
     echo "  --help, -h              Show this help message"
     echo "  --list, -l              List all available skills"
     echo "  --all, -a               Install all skills (default if no skills specified)"
-    echo "  --local                 Install from local files instead of downloading"
+    echo "  --local                 Install from the frozen local copies under DEPRECATED-databricks-skills/ instead of downloading"
     echo "  --install-to-genie      After install, upload ./.claude/skills to workspace"
     echo "                          /Users/<you>/.assistant/skills for Genie Code (uses cwd; requires databricks CLI)"
     echo "  --profile <name>        Databricks config profile for workspace upload (default: DEFAULT or \$DATABRICKS_CONFIG_PROFILE)"
@@ -312,9 +326,9 @@ download_databricks_skill() {
     local skill_dir="$SKILLS_DIR/$skill_name"
 
     if [ "$INSTALL_FROM_LOCAL" = true ]; then
-        # Copy from local files
-        echo -e "  Copying from local..."
-        local source_dir="$SCRIPT_DIR/${skill_name}"
+        # Copy from the frozen legacy copies under DEPRECATED-databricks-skills/
+        echo -e "  Copying from local (DEPRECATED-databricks-skills/)..."
+        local source_dir="$SCRIPT_DIR/../DEPRECATED-databricks-skills/${skill_name}"
 
         # Check if source directory exists
         if [ ! -d "$source_dir" ]; then
@@ -349,8 +363,8 @@ download_databricks_skill() {
             done
         fi
     else
-        # Download from URL
-        echo -e "  Downloading from Databricks repo..."
+        # Download from URL (pinned to the frozen ${REPO_REF} tag)
+        echo -e "  Downloading from Databricks repo (${REPO_REF})..."
 
         # Download SKILL.md (required)
         if curl -sSL -f "${REPO_RAW_URL}/databricks-skills/${skill_name}/SKILL.md" -o "$skill_dir/SKILL.md" 2>/dev/null; then
@@ -525,6 +539,25 @@ echo -e "${BLUE}║        Databricks Skills Installer for Claude Code         
 echo -e "${BLUE}╚════════════════════════════════════════════════════════════╝${NC}"
 echo ""
 
+# Deprecation banner
+echo -e "${YELLOW}────────────────────────────────────────────────────────────${NC}"
+echo -e "${YELLOW}  ⚠️  DEPRECATED SCRIPT${NC}"
+echo -e "${YELLOW}────────────────────────────────────────────────────────────${NC}"
+echo -e "  The supported way to install editor skills is now:"
+echo -e "    ${GREEN}databricks aitools install${NC}   (Databricks CLI v1.0.0+)"
+echo -e "  or the AI Dev Kit installer, which delegates to it:"
+echo -e "    ${GREEN}bash <(curl -sL https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/main/install.sh)${NC}"
+echo ""
+echo -e "  This script is kept only for the ${GREEN}--install-to-genie${NC} workspace"
+echo -e "  upload flow. For that, the notebook uploader is simpler:"
+echo -e "    ${GREEN}databricks-skills/install_genie_code_skills.py${NC} (no local clone)"
+echo ""
+echo -e "  Skill files are ${YELLOW}frozen legacy copies${NC}, no longer maintained:"
+echo -e "  the default downloads them from release tag ${REPO_REF}; ${GREEN}--local${NC} uses"
+echo -e "  the DEPRECATED-databricks-skills/ folder instead."
+echo -e "${YELLOW}────────────────────────────────────────────────────────────${NC}"
+echo ""
+
 # Check if we're in a git repo or project directory
 if [ ! -d ".git" ] && [ ! -f "pyproject.toml" ] && [ ! -f "package.json" ] && [ ! -f "databricks.yml" ]; then
     echo -e "${YELLOW}Warning: This doesn't look like a project root directory.${NC}"