agentleFS
kjranyonekjranyone/artcraft-agent-skills/skills/effectcraft/SKILL.md

EffectCraft

Skill1 starsChanged 17 days agoMITView on GitHub

Use whenever working with EffectCraft on this machine — the Rust motion-graphics / VFX compositor and After Effects alternative — including launching the app with control port ("EffectCraftを起動", "EffectCraftを立ち上げて", "launch EffectCraft"), creating or editing compositions, layers, keyframes, masks, tracking, rotoscoping, effects and text (.ecproj, MP4/MOV/WebM/ProRes/EXR/PNG sequences, .ectemplate, .ecpreset, ExtendScript/.jsx), driving the running desktop window, or using its MCP server, effectcraft-cli or control channel. Triggers on "EffectCraft", "effectcraft-cli", "effectcraft mcp", "EffectCraft MCP", "EffectCraftを起動", "EffectCraft立ち上げて", "EffectCraft開いて", "make a motion graphic", "animate this text", "add an effect", "track this shot", "rotoscope", "export to MP4/ProRes/EXR", "take an EffectCraft screenshot", and requests to automate motion graphics, VFX or EffectCraft itself. Invoke with @effectcraft.

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. midasmatterhorn/artcraft-skills★ 82same name
    2. hermes-pro/storytold-skills★ 36same 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 repository4 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 4 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Check that EffectCraft CLI is installed and compatible:

    line 15
  2. Distinctive:

    If command not found: tell the user "EffectCraft is not installed.

    line 21
  3. Distinctive:

    If version < 0.17: warn about API differences (see bugs-and-workarounds.md)

    line 22
  4. Distinctive:

    If version ≥ 0.18: check bugs-and-workarounds.md for breaking changes

    line 23
  5. Very specific:

    Every layer has a property tree: Transform (anchor, position, scale, rotation, opacity), plus effects, masks, text or shape contents.

    line 47
  6. Very specific:

    If running, test the port: effectcraft-cli --bridge 50052 info.

    line 76
  7. Very specific:

    If running *without* a control port, ask the user to restart it with --control (unsaved work would be lost otherwise).

    line 77
  8. Very specific:

    Verify: wait 1–3 s, then effectcraft-cli --bridge 50052 info — it prints the version, command/effect counts and the open project.

    line 84
  9. Very specific:

    Start with: comp.new, comp.settings, comp.open, layer.newText, layer.newShape, layer.newSolid, layer.newNull, layer.select, …

    line 143
  10. Very specific:

    Layer time vs comp time: Write commands (prop.addKey, keys.set, prop.set) take time in layer seconds by default (timeBase:"layer"|"comp" …

    line 157

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

EffectCraft — agent operation

EffectCraft is a native Rust motion-graphics / VFX compositor (After Effects-class) whose agent interface is built in: every action is a command with an id. The GUI, menus, the CLI, the control channel and the MCP server all drive one engine. Verified against 0.17.0 on Windows.

Setup (agent: run before first use)

Check that EffectCraft CLI is installed and compatible:

effectcraft-cli --version   # expect 0.17.x
  • If command not found: tell the user "EffectCraft is not installed. Download from https://getartcraft.com/apps/effectcraft"
  • If version < 0.17: warn about API differences (see bugs-and-workarounds.md)
  • If version ≥ 0.18: check bugs-and-workarounds.md for breaking changes

Machine & environment paths

PlatformBinaryCLIDefault Data Dir
WindowsC:\Program Files\EffectCraft\effectcraft.exeC:\Program Files\EffectCraft\effectcraft-cli.exe%LOCALAPPDATA%\EffectCraft (Disk Cache)
Linux/usr/bin/effectcraft/usr/bin/effectcraft-cli~/.local/share/effectcraft
  • Control port: --control <port> or EFFECTCRAFT_CONTROL_PORT (no token — the bridge accepts plain loopback connections, unlike FilmCraft)
  • Commands: 734 in 55 namespaces (439 enabled in a fresh headless context; the rest need a selection / active comp / tool mode)
  • Effects: 307 in the registry (list_effects); 149 font families (list_fonts)

Mental model

  • A project (.ecproj) holds items: Composition, Folder, Solid, Footage. get_project returns the item tree, the dirty flag and the active comp. One comp is active; most commands act on it.
  • A comp is {width, height, frameRate, duration, startTime, background, renderer: classic3D|advanced3D}. Times are seconds (floats, not ticks — no tick math here, unlike FilmCraft).
  • A layer is Text, Shape, Solid, Null, Composition (precomp), Adjustment, Light or Camera. Layers stack 1 = top. Address layers by id, #n (stack index) or name.
  • Every layer has a property tree: Transform (anchor, position, scale, rotation, opacity), plus effects, masks, text or shape contents. Property paths accept transform.position and Transform/Position — but a bare position (no group) is rejected.
  • Animation = keyframes on properties (prop.addKey) or expressions (prop.setExpression, AE JavaScript). prop.get {time} returns the interpolated value, the key list and the animated flag.
  • Parameters are validated strictly: an unknown parameter name fails the command with the accepted-key list; so does an unknown command id. This is the opposite of FilmCraft's silently-dropped params — read the error, fix the call, don't retry blind.
  • Verify by reading back (get_state, get_project, get_comp, get_layer, get_property) or by rendering (render_frame — a PNG catches wrong positions, invisible layers and missing effects that no read-back shows).

Two connection modes

ModeStartUse for
Headlesseffectcraft-cli mcp --demo (--project p.ecproj, --empty, --autosave)no window, batch builds, renders, scripts
App bridgeeffectcraft --control 50052 + effectcraft-cli mcp --bridge 50052the real window: UI state, screenshots, clicks, keys, GUI-only commands

Each CLI invocation is its own session. State built in one run is gone when the process exits — finish with file.saveAs + --save-as, or do everything (build → render) in a single render/run invocation.

Launching EffectCraft on user request ("起動して" / "start EffectCraft")

  1. Check if already running: Get-Process effectcraft / pgrep effectcraft. If running, test the port: effectcraft-cli --bridge 50052 info. If running without a control port, ask the user to restart it with --control (unsaved work would be lost otherwise).
  2. Launch with --control <port> (default 50052):
    • Windows — use scripts/launch.ps1 so the window appears on the visible desktop instead of a background session: powershell -File scripts/launch.ps1 -Port 50052
    • Linux: ./scripts/launch.sh 50052
  3. Verify: wait 1–3 s, then effectcraft-cli --bridge 50052 info — it prints the version, command/effect counts and the open project.

The MCP tools

Headless exposes 27 tools; the bridge adds 9 UI tools (ui_inspect, ui_elements, ui_click, ui_drag, ui_key, ui_set, ui_type, screenshot, control) plus script_ui for ScriptUI dialogs = 36.

ToolPurpose
command_list / list_commandsevery command (--filter, --enabled)
command_run / execute_commandrun one {id, params} (undoable)
command_batch / batchseveral commands, one undo step (batch)
describe_commandone command's schema
doc_inspect / get_projectproject tree + active comp in one call
get_compone comp: settings, layers, work area
get_layera layer's full property tree
get_property / set_propertyread / write one property (set keys when animated)
add_keyframekeyframes for a property (turns the stopwatch on)
add_effect / list_effectsapply an effect (+params in one call) / the 307-effect registry
list_fonts149 font families usable by text layers
run_scriptAfter Effects-style ExtendScript (app.project, comps, layers)
render_frame / render_previewPNG of the (active) comp at time
open_project / save_projectopen (path, demo:true, empty:true) / save (+Save As)
undo / redo / historybranching undo history
get_stateactive comp, current time, selection, playback
script_uidrive ScriptUI dialogs/palettes scripts opened

Resources: effectcraft://document and effectcraft://commands.

Standard workflow

  1. Probe. get_state + get_project; headless starts empty unless --demo/--project. ui_inspect first in bridge mode.
  2. Build. comp.new {name,width,height,frameRate,duration} → layer.newText/newShape/newSolid/newNull (styling params inline) → file.import {paths} for footage → layer.addItem-style placement.
  3. Animate. prop.addKey {layer,path,time,value} per key; keys.setEase, keys.interpolation, prop.setExpression.
  4. Style. effect.apply {effect:"Gaussian Blur", layers:[id]} + param paths under effects/#1/...; masks (mask.new), text animators, shape gradient strokes.
  5. Verify. get_property at telling times; render_frame at those times and look at the PNGs (position, easing and blur are all visible).
  6. Render. render --out file.mp4 --format h264 (CLI), or renderQueue.add + renderQueue.render; render-frame --transparent for RGBA; render --template t.ectemplate --values v.json for template jobs.
  7. Save. file.saveAs {path} (or CLI --save-as); re-read get_project — the dirty flag clears after save.

Finding commands

command_list {filter:"mask"} matches id, label and menu. From a shell: effectcraft-cli commands --filter mask, and describe_command/describe for one command's schema. All 734 ids with parameter docs are in references/catalog/commands.md; verified step-by-step JSON is in references/recipes.md.

Start with: comp.new, comp.settings, comp.open, layer.newText, layer.newShape, layer.newSolid, layer.newNull, layer.select, layer.moveLater, layer.setTransform, layer.setText, prop.get, prop.set, prop.addKey, keys.setEase, keys.interpolation, prop.setExpression, effect.apply, mask.new, time.set, file.import, file.saveAs, renderQueue.add, renderQueue.render, app.quit.

Pitfalls that cost time

  1. Strict validation everywhere. Unknown param → invalid parameters … accepted: name, width, … with the full key list; unknown command id → unknown command. Errors are the documentation — but a valid param with a wrong value type can still misbehave, so read state back.
  2. Layer time vs comp time. Write commands (prop.addKey, keys.set, prop.set) take time in layer seconds by default (timeBase:"layer"|"comp" switches); prop.get reads time in comp seconds. With a layer starting at 0 they coincide; after layer.move they don't. Wrong base = keys silently offset by the layer start.
  3. Many commands need a selection. prop.set, effect.apply, keys.easyEase show enabled:false with nothing selected (headless starts empty). Pass explicit layer/layers params, or layer.select first.
  4. comp.new returns {comp: id} and makes it the active comp (verified: two comp.new calls, a text layer lands in the second). Switch with comp.open {comp: id|name} and confirm via get_state — commands and renders that target "the active comp" fail with no active composition on a fresh --empty project before the first comp.new.
  5. Property paths need the group: transform.position / Transform/Position work, bare position errors with no property. Find exact paths with props <comp> <layer> --flat or get_layer.
  6. Effect params are paths under the effect, e.g. effects/#1/Blurriness (index = application order). effect.apply returns paths:["effects/#1"] — use them.
  7. Each CLI invocation is a fresh session (see above). MCP sessions keep state until the server exits.
  8. render writes relative --out to the process working directory; always pass an absolute path for predictable output.
  9. Text layers take full styling at creation (layer.newText {size,fill,tracking,…} — dozens of params). Later restyling goes through text source/document properties; check get_layer for the exact paths.
  10. GUI-only surface: ui_*, screenshot, control, script_ui exist only on the bridge; window.*, view.* and help.* mostly need the GUI.
  11. mcp --autosave persists unsaved headless edits on exit; without it they are lost (expected for throwaway runs, surprising for real work).

Reference map

FileContents
references/recipes.mdverified JSON: build, text, shape, keyframes, easing, expressions, effects, masks, render, export, GUI
references/catalog/commands.mdall 734 commands grouped by namespace with parameter docs
references/mcp-and-control.mdthe 27+9 tools, error shapes, resources, bridge details
references/cli.mdeffectcraft-cli subcommands, render formats, templates, bench
references/bugs-and-workarounds.mdreproduced defects with minimal reproductions
references/maintenance.mdre-verify the version, dump commands, refresh this skill
scripts/probe.sh / .ps1MCP handshake + tool count + version
scripts/dump-commands.sh / .ps1regenerate the command catalog
scripts/launch.sh / .ps1start the GUI with a control port (visible desktop)
scripts/smoke-test.sh / .ps1build → keyframe → render → validate

Community: https://discord.gg/artcraft · https://getartcraft.com/apps/effectcraft.

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.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

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.