메뉴
Deploy, configure, and operate HAL Plus — the local AI-powered web UI layer for HAL. Use when the user asks to start HAL Plus, switch models, check its health, or understand how it connects to Ollama and HAL MCP.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Deploy the local Terraform demo environment in hal. Use to provision and initialize Terraform lab resources.
Route Terraform lab requests to deploy, destroy, inspect status, and manage workspace automation in hal.
Deploy local Vault in hal. Use when the user asks to start Vault, initialize the local control plane, or recover from Vault being offline.
Route user intent to the right Vault lab skill in hal, including CE versus Enterprise constraints, observability wiring, and full-stack scenario guidance.
Deploy, verify, and troubleshoot the Vault JWT auth lab in hal. Use this skill when the user asks to enable JWT auth, configure CI/CD authentication with Vault, debug bound claims, test GitLab or pipeline logins, or reset the JWT demo. Triggers include "enable jwt", "configure jwt", "gitlab vault auth", "pipeline auth", "bound claims", and "hal vault jwt".
Deploy, verify, and troubleshoot the Vault OIDC lab in hal. Use this skill when the user asks to enable OIDC, set up human SSO, debug browser login redirects, inspect Authentik-backed roles, reset the OIDC demo, or configure SCIM provisioning. Triggers include "enable oidc", "Vault SSO", "Authentik", "OIDC callback", "configure oidc", "hal vault oidc", "enable scim", and "vault scim".
| name | plus |
| description | Deploy, configure, and operate HAL Plus — the local AI-powered web UI layer for HAL. Use when the user asks to start HAL Plus, switch models, check its health, or understand how it connects to Ollama and HAL MCP. |
HAL Plus is the local web UI for HAL. It runs as a container alongside HAL MCP and proxies queries to an Ollama model on the host. Answers are grounded through the HAL MCP runtime tools.
ollama serve) before hal plus create.host.containers.internal:11434 from inside the container.hal mcp serve) runs alongside HAL Plus on the same container network.qwen3.5 (32k context window).hal plus create
hal plus status
hal plus delete
hal plus create --model qwen3.5 # default — balanced, 32k context
hal plus create --model gemma4 # lighter, faster cold starts
Both presets are HAL-managed — no manual ollama pull needed.
hal plus create --model qwen3.5 --keep-alive 5m
hal plus create --model qwen3.5 --model-config ./Modelfile
hal plus create --image localhost/hal-plus:testmiche
Use a locally built image (e.g. during development or testing a branch).
| Service | URL |
|---|---|
| HAL Plus UI | http://hal.localhost:9000 |
| HAL Plus UI (dark mode) | http://hal.localhost:9000/dark |
| HAL Plus API | http://hal.localhost:9000/api |
http://127.0.0.1:18080/mcp from inside HAL Plus.LLM offline, ensure Ollama is running (ollama serve) and the model is present (ollama list).hal mcp status to confirm the MCP server is reachable.hal plus create fails with an image pull error, either push the image to a registry or use --image localhost/hal-plus:local with a locally built image.hal plus delete removes the container but does NOT unload the Ollama model from the host.