Deploy Grokky
SkillCloud & infraDeploy Grokky — start Docker, ensure grok_spawner, build images, publish package
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Deploy Grokky skill
What this skill tells your AI
The instructions your AI receives, as published by datagrok-ai/public in packages/Grokky/.claude/skills/deploy-grokky/SKILL.md and read by ahel’s review.
This skill performs a full deployment of the Grokky package:
- Ensures Docker is running
- Ensures grok_spawner is running
- Builds required Docker images
- Starts runtime containers
- Builds and publishes the package
Arguments
$ARGUMENTS may include:
key: <ANTHROPIC_API_KEY>host: <publish host>
Resolve arguments
-
If
hostis missing: Ask: "Which host should I publish to? (e.g.,localhost,dev,public)" Wait for user input before continuing. -
If
keyis missing: Run:echo $ANTHROPIC_API_KEYIf empty, ask the user to provide the key and wait.
Resolve paths
Set variables explicitly:
GROKKY_ROOT=$(cd "${CLAUDE_SKILL_DIR}/../../.." && pwd)
DOCKER_DIR=$(cd "$GROKKY_ROOT/../../docker" && pwd)
COMPOSE_FILE="$DOCKER_DIR/localhost.bleeding-edge.docker-compose.yaml"
Step 1 — Ensure Docker is running
Run:
docker info > /dev/null 2>&1
If it fails:
open -a Docker
Then retry every 3 seconds (max 10 attempts):
for i in {1..10}; do
sleep 3
docker info > /dev/null 2>&1 && break
done
If still failing after 10 attempts: Stop execution and tell user: "Docker is not running. Please start Docker manually."
Step 2 — Ensure grok_spawner is running and fresh
Check if running:
docker ps --filter "name=grok_spawner" --format "{{.Names}}"
If no output, start it:
cd "$DOCKER_DIR"
GROK_SPAWNER_CORE_MODE=true docker compose -f "$COMPOSE_FILE" up -d grok_spawner
If running, check its age:
docker inspect --format '{{.Created}}' grok_spawner
Parse the timestamp and compare to now. If the container is older than 2 days, recreate it:
cd "$DOCKER_DIR"
GROK_SPAWNER_CORE_MODE=true docker compose -f "$COMPOSE_FILE" up -d --force-recreate grok_spawner
Do NOT start any other services.
Step 3 — Detect the Docker network
The network name varies by setup. Detect it from the running grok_spawner container:
DOCKER_NETWORK=$(docker ps --filter "name=grok_spawner" --format "{{.Networks}}" | head -1)
If DOCKER_NETWORK is empty, fall back:
DOCKER_NETWORK=$(docker network ls --format "{{.Name}}" | grep -E "datagrok" | head -1)
If still empty, stop and tell user: "No datagrok Docker network found. Is the platform running?"
Step 4 — Detect required image builds
cd "$GROKKY_ROOT"
CHANGED_FILES=$(git diff --name-only HEAD -- dockerfiles/; git status --short -- dockerfiles/)
Check existing images:
EXISTING_IMAGES=$(docker images --format "{{.Repository}}:{{.Tag}}" | grep grokky || true)
Determine build flags:
-
Build
claude-runtimeif:dockerfiles/claude-runtime/appears in$CHANGED_FILESORgrokky-claude-runtime:adminnot found in$EXISTING_IMAGES
-
Build
mcp-serverif:dockerfiles/mcp-server/appears in$CHANGED_FILESORgrokky-mcp-server:adminnot found in$EXISTING_IMAGES
If neither needs building, skip to Step 6.
Step 5 — Build images (only those flagged in Step 4)
claude-runtime
cd "$GROKKY_ROOT/dockerfiles/claude-runtime"
docker build --build-arg ANTHROPIC_API_KEY=$KEY -t grokky-claude-runtime:admin .
docker rm -f grokky-claude-runtime 2>/dev/null || true
mcp-server
cd "$GROKKY_ROOT/dockerfiles/mcp-server"
docker build -t grokky-mcp-server:admin .
docker rm -f grokky-mcp-server 2>/dev/null || true
Step 6 — Build and publish
cd "$GROKKY_ROOT"
npm run build
If build fails, stop and show the error output.
If build succeeds:
grok publish $HOST
If publish fails, stop and show the error output.
Step 7 — Report results
Print:
- Whether Docker was started
- Whether grok_spawner was started
- Which images were built (or "all cached")
- Container statuses
- Publish result
Signals
- GitHub stars
- 72
- Forks
- 32
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
deploy-grokky- Source
- github.com/datagrok-ai/public