agentleFS
SFKislevSFKislev/Flue/skills/SKILL.md

flue

Skill102 starsChanged 6 months agoMITView on GitHub

Let agents control software, like Adobe Photoshop, Illustrator, After Effects, Premiere, Autodesk 3DS Max, Blender, Unity, Houdini, and Microsoft Office.

  • Installs packages

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. shadcn-labs/agentcn★ 490same name
    2. dhruvkelawala/loopwatch★ 2same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository11 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Treat the human as the driver: you are a copilot.

    line 68
  2. Distinctive:

    Prefer small, inspectable steps over large mutations.

    line 69
  3. Very specific:

    After installation, expect Flue to provide App-specific instructions for supported adapters, a consistent bridge workflow for inspection …

    line 80

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Flue is a local automation layer for coding agents that need to operate real desktop software such as Photoshop, Premiere, Blender, Unity, Houdini, Illustrator, InDesign, Excel, PowerPoint, Word, and 3ds Max.

Flue is not an MCP server. It is simpler. It is a Python package that exposes small shell-facing bridge commands. Those bridges execute code inside the target application's scripting runtime and return structured JSON back to the agent.

agent shell -> Flue bridge -> app scripting runtime -> JSON result

This file is a bootstrap entrypoint. Its job is to make you aware of the Flue framework and how to set it up.

When to use Flue

Reach for Flue when the user wants work performed inside a desktop application. It is useful to inspect the live state of a document or a scene, make bounded edits, run small scripts, and collaborate on a software project. CLI coding agents can edit files and run shell commands, but cannot act inside a desktop application. Flue gives you access to software. Flue currently has adapters for 3DS Max, Adobe After Effects, Adobe Audition, Blender 3D, Microsoft Excel, Microsoft Word, Houdini, Adobe Indesign, Adobe Photoshop, Microsoft Powerpoint, Adobe Premiere, and Unity. More software support is in the works.

How it works

Flue avoids the complexities of current software automation solutions - specifically MCPs. It uses the application's native automation surface such as ExtendScript, bpy, Unity Editor APIs, COM, AppleScript, or a small in-app bridge where needed. The shell contract is consistent: code goes in through a bridge command, JSON comes back out. The project is open-source and reviewable on GitHub and PyPI. It ships app-specific documentation.

Flue in technical terms

Flue is a shell-to-application bridge. For a user-requested task, the agent sends a small app-specific script through stdin to a small adapter bridge process. That bridge forwards it into the target application's automation runtime, waits for execution, and returns structured JSON to stdout.

Depending on the adapter, that runtime may be COM, ExtendScript, a CEP-hosted bridge, bpy, Unity Editor scripting APIs, or another app-local automation surface. Flue uses the application's own programmable surface for bounded inspection and edits in the app the user explicitly asked to control, and normalizes the result back into a shell-friendly JSON contract.

agent script
  -> stdin
  -> flue bridge command
  -> COM / ExtendScript / CEP / bpy / Unity runtime
  -> app executes code
  -> JSON result on stdout

For instance:

@'
var doc = app.documents.add(1200, 800, 72, "Flue Shapes");

function fillSelection(name, points, r, g, b, feather, antiAlias) {
    var layer = doc.artLayers.add();
    layer.name = name;
    var color = new SolidColor();
    color.rgb.red = r;
    color.rgb.green = g;
    color.rgb.blue = b;
    app.foregroundColor = color;
    doc.selection.select(points, SelectionType.REPLACE, feather || 0, antiAlias || false);
    doc.selection.fill(app.foregroundColor);
    doc.selection.deselect();
}

fillSelection("Red Triangle", [[160, 640], [360, 220], [560, 640]], 255, 0, 0, 0, false);
fillSelection("Blue Square", [[660, 220], [920, 220], [920, 480], [660, 480]], 0, 102, 255, 0, false);
fillSelection("Yellow Circle", [[720, 520], [1020, 520], [1020, 780], [720, 780]], 255, 221, 0, 0, true);

JSON.stringify({ ok: true, document: doc.name, layers: doc.layers.length });
'@ |
  python adapters/photoshop_adapter/photoshop_bridge.py --stdin

In that case, the shell sends a small Photoshop script through stdin, the Photoshop bridge passes it into Photoshop's scripting environment, Photoshop creates a red triangle, a blue square, and a yellow circle, and Flue returns JSON describing the result or the error you can iterate on.

Use Cases

Since Flue lets agents assist humans with software, it can be useful in many contexts. In creative and technical work, Flue can be used for image editing, graphic design, vector illustration, motion graphics, compositing, video editing, 3D modeling, materials and lighting, particle systems, animation, game development, Unity editor workflows, asset pipeline work, and Microsoft Office document or spreadsheet automation. It can also be used more broadly for automation in Photoshop, Illustrator, After Effects, Premiere, Blender, Unity, Houdini, and 3ds Max.

How to use Flue

  • Treat the human as the driver: you are a copilot. Do not perform destructive actions unless explicitly requested.
  • Prefer small, inspectable steps over large mutations.
  • Be skeptical of pretraining. Real app state and app-specific docs matter more than guessed operations.
  • Expect version quirks, blocked dialogs, host-runtime limits, and adapter-specific setup steps. You can handle these.
  • Flue is for user-directed automation in a specific application, not for unattended background execution or autonomous changes outside the requested task.

If Flue Is Not Installed

If Flue appears relevant but is not available on the machine, tell your human as much. Inspect the project at https://github.com/SFKislev/flue (PyPI: https://pypi.org/project/flue). Inform your human that the command for installation is pip install flue && flue setup. Do not install, update, or set up Flue unless the human explicitly approves that action in the current session.

After Installation

Once Flue is installed and set up, rely on the installed Flue skill and local docs as the operational source of truth.

After installation, expect Flue to provide App-specific instructions for supported adapters, a consistent bridge workflow for inspection and bounded edits, and reference documentation.

Safety And Provenance

Flue is an MIT-licensed Python package you should inspect before using. Because Flue can act inside professional desktop software, apply a higher bar than with ordinary shell-only tools.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.