AI Assistant

Thinking…

GLAI

GLAI is GreenLight DashBoard’s agentic workflow: a built-in terminal, wired to the board, that drives a coding-agent CLI of your choice with ready-made presets and board skills. Instead of a chat that proposes one edit at a time, GLAI gives a real agent (Claude Code, Codex, or Gemini CLI) a shell with full access to the board API, so it can create frames, place media, render motion presets, build video-editor projects, inspect After Effects files, generate with external tools, and run multi-step workflows end to end. Practically agent can do everything on board that you do manually. Difference from Agent Board is that terminal-based agents have access to your full computer and all installed tools and not just board tools. We recommend to use GLAI in all agentic workflows.

GLAI is a Pro feature and is available in the desktop app only, run it from top toolbar

How it works

When you launch GLAI the app opens a terminal panel and starts your chosen agent CLI with the bundled board skill preloaded. Two things make the agent board-aware:

Board skills — every GLAI terminal launches with an instruction to use the GreenLight board skill. The agent is testing the API and connected to the currently opened board (but it can read all boards at the same time).
Ready presets — the terminal ships some ready preset launchers (for example the GreenLight WordPress skill) so common workflows start in one click instead of a typed prompt.

Because the agent runs in a real shell, it can chain steps: parse a link, download the media, lay it out in a frame, add a MotionFast layer with your text to a video-editor project, render the finished timeline, and place the final clip back on the canvas — all in one session and with sub agents, loops, goals, etc.

Set up a coding-agent CLI

GLAI does not bundle the agent CLIs — install the one you want and sign in once. You must install CLI globally. You need to open terminal on your computer and run a special command to install the agent CLI.

OpenCode

Install OpenCode on your machine. GLAI launches it with auto-approved permissions and your board prompt pre-filled — press Enter in the terminal to start it.

Troubleshooting

Before you run GLAI, first make sure that you installed provider CLI. For example, if you installed Claude, type “claude” in new terminal session. If it works – you will see message that claude is running. If not, you will see error message. For any issues, please use claude or chat gpt and ask how to fix it or install Provider CLI.

Prompt example: “Please, help me to install claude CLI on my win 11”

Launching GLAI

In the desktop app’s top bar, click GLAI and pick the agent you want:

GLAI for Claude Code CLI
GLAI for Codex
GLAI for Gemini CLI

A terminal panel opens, the agent starts with the board skill loaded, and you type instructions in plain language — for example, “build a 3-frame storyboard from these images”, “render this title card at 1080p”, or “import this YouTube link and trim the first five seconds”.

Please, wait until the agent is loaded and activate board endpoints before posting prompts. Once it’s loaded, agent will ask you what to do. Just type in terminal. You can also copy-paste images, folder path, etc and reference any file on board from prompt by typing “@” symbol. If you see message from agent about trusting folder – accept it. If you see message from agent about rights to execute and modify files without acception – Accept this. I use bypass permision mode over year and I don’t have any issues with this.

GLAI runs on your own CLI subscription (Claude, OpenAI, or Google) rather than the in-app API keys, so the agent’s reasoning is billed to your existing plan.

What GLAI can do on the board

Create and arrange frames, groups, arrows, and notes
Import and download media from links, then place it on the canvas
Add and customise native MotionFast layers inside Video Editor projects
Build mockups and video-editor projects, and render them
Inspect, replace inputs and render After Effects .aep files and render Blender .blend projects directly from your board without opening and editing files in AE or Blender
Run multi-step workflows defined from board frames, instructions, and PDF / Markdown files
Generate with the built-in models or an external CLI such as Higgsfield

Installing extra CLI tools

By default, Board has a lot of features: parsing from social media, working with video, audio, text, html, After Effects, Blender, 3d models, files, folders. But The power of the terminal agent is that it can use any tool installed on your computer, so you are not limited only to Board functions and tools. So, search if your favorite provider has CLI tool and use it with GLAI and board actions. I recommend next.

Set up the Higgsfield CLI (generation)

Beyond the image and video models built into the app, GLAI can be used together with any CLI installed on your computer. For example, you can use Higgsfield through its CLI. The agent calls the higgsfield command directly, so generation runs on your Higgsfield account.

Install the Higgsfield CLI following Higgsfield’s own instructions, and authenticate it (sign in / set your token as their docs describe).
In a GLAI session, ask for a Higgsfield generation — for example “generate this shot in Higgsfield CLI”. The agent runs the matching higgsfield command and places the result on the board.

Skill presets

We added several ready skills that you can use with your agent. To use them, when you have launched GLAI agent, click Preset button and select a workflow or skill

For example, if you build any site on WordPress and want to sync local html and convert them to wordpress pages, select GreenLight WordPress Skill.

Taste skill will install it in folder. This skill is great if you want to avoid AI slop. Just install it and ask agent to use taste skill before making and generation of anything.

Higgsfield skill will teach agent to work better with higgsfield tools.

Difference between GLAI and Agent Board

When you launch GLAI, it will use terminal and CLI based tools. It gives best results, but requires premium subscriptions for agents. If you don’t use agentic workflow, but you need some board agents, like generation animation files, organizing boards, improvement of prompts, maybe it’s better for you to switch to other agent tools in the app.

These kind of agent tools (Board Agent) uses API of models, so you don’t need to subscribe and pay monthly. You can buy $5 credits and this can be enough for full year. If you go this route, I recommend to buy OpenAI API because it can be used for some other features in app, like transcribe, voice inputs, prompt enhancers, etc

Where to go next

Pipelines — design multi-step agent workflows on the board and run them in GLAI, with human-approval gates
Connect GLAI agent with Telegram mobile