agentsclimarketplace

AI SKILL gimp

Skill LevyBytes/AI-SKILL-gimp

Agent Kaizen skill for Claude Code & Codex: faithful GIMP 3.2 reference (GNU Image Manipulation Program), layers, blend modes, masks, channels, selections, Bézier paths, levels, curves, ICC profiles, .xcf/PNG/JPEG export, and scripting via Script-Fu, Python-Fu & the PDB, including headless batch and GIMP MCP server runs.

Install
npx -y skills add LevyBytes/AI-SKILL-gimp

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when working with GIMP, the GNU Image Manipulation Program (especially GIMP 3.2) - editing or generating raster images; working with layers, layer/blend modes, masks, channels, selections, or Bézier paths; applying tools or filters; adjusting colors, levels, curves, or ICC color profiles; importing or exporting image files (.xcf, PNG, JPEG); or automating GIMP through Script-Fu, Python-Fu, the Script-Fu Console, or the Procedural Database (PDB) - including driving GIMP from a GIMP MCP server or a headless/batch (gimp -i -b) run. Also fires when you see .xcf files, gimp-* procedure calls, script-fu-register, "from gi.repository import Gimp", or GObject-Introspection GIMP plug-ins, even when the user never says "GIMP" by name. Gives task-routed, faithful 3.2 reference with worked examples.

SKILL.md

6.1 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it

GIMP 3.2

Task-organized, faithful guidance for GIMP 3.2 (GNU Image Manipulation Program), with emphasis on driving GIMP programmatically through Script-Fu, Python-Fu, and the Procedural Database (PDB) — for example from a GIMP MCP server. Content is original; identifiers (menu paths, option labels, keyboard shortcuts, mode names, procedure names, numbers) are preserved exactly as the manual prints them.

When to use this skill

Reach for this skill whenever a task touches GIMP: editing or generating images; layers, blend modes, masks, channels, selections, or paths; tools or filters; color adjustments and color management; opening, saving (.xcf), or exporting (PNG/JPEG); or automating GIMP via Script-Fu / Python-Fu / the PDB — including when the user opens a .xcf file, pastes gimp-* calls, writes script-fu-register or from gi.repository import Gimp, or wires up a GIMP MCP server, without naming "GIMP".

How to use it (runtime workflow)

  1. Read this file, then consult the task router below (or references/topics.json).
  2. Open only the one or two reference files relevant to the task — do not read everything (progressive disclosure).
  3. Answer strictly from the reference; preserve menu paths, option labels, shortcuts, mode names, and procedure names exactly as written.
  4. If something is not covered, say so and mark it TODO — never invent a menu path, option, or PDB procedure name. For scripting, the in-app Help > Procedure Browser is the authoritative source for exact procedure names, parameters, and return types.
  5. Remember the GIMP 3.x API is not 2.10: Python-Fu is Python 3 + GObject Introspection, and many PDB procedures were renamed — verify names before relying on them.

Task router

If the task is about…Read
windows, docks/dialogs, the menu map, zoom/undo, image Mode or Encoding (bit depth)references/interface-and-workflow.md
the Toolbox, Tool Options, which tool to use and its shortcutreferences/tools-overview.md
layers, the Layers dialog, groups, alpha, opacity, blend/paint modesreferences/layers-and-modes.md
selections, selection modes, feathering, the Select menureferences/selections.md
layer masks, the Channels dialog, saving/loading selections, Quick Maskreferences/masks-and-channels.md
Bézier paths, the Paths dialog, stroking, path↔selection, SVGreferences/paths.md
the Colors menu, levels/curves/hue-saturation, ICC profiles (assign vs convert)references/colors-and-adjustments.md
the Filters menu taxonomy, Repeat/Re-Show Lastreferences/filters.md
opening/importing, saving .xcf, exporting/overwriting, PNG/JPEG optionsreferences/file-formats-and-export.md
Script-Fu vs Python-Fu, the consoles, install dirs, batch/headless flagsreferences/scripting-overview.md
the PDB, Procedure Browser, procedure naming and typesreferences/pdb-reference.md
Scheme syntax, script-fu-register / script-fu-menu-registerreferences/script-fu.md
Python 3 + GObject Introspection, the Gimp.PlugIn modelreferences/python-fu.md
scripting image/layer ops (create, fill, resize, flatten)references/pdb-recipes-images-layers.md
scripting selections and pathsreferences/pdb-recipes-selections-paths.md
scripting color adjustments and filtersreferences/pdb-recipes-color-filters.md
scripting file save/export and headless batch runsreferences/pdb-recipes-io-batch.md

Gotchas

Recurring failure modes and what to do instead live in the sibling GOTCHA.md.

Verification notes

Validate the package structure (the validator ships with the skill-drafting skill; paths use the DEVROOT env var — adjust to your layout):

python "$env:DEVROOT\SKILLS\skills\skill-drafting\scripts\validate_skill_package.py" "$env:DEVROOT\SKILLS\skills\gimp"

This package is built and checked with the three-layer method (Spec → Verifier → Environment). Ground-truth checks live in AI/work/verify_gimp.py (stdlib only; run python AI/work/verify_gimp.py --both). It verifies: every reference is listed in references/INDEX.md, has exactly one references/topics.json entry, and resolves its cross-links; each reference contains a summary, a "When this applies" line, at least one example, and a Do/Don't or anti-patterns section, with no empty placeholders; an originality gate rejects source branding/attribution/license leakage in the references and in the SKILL.md body (outside "Inspired by"); a preserve-glossary of load-bearing identifiers (menu paths, shortcuts, mode names, PDB procedure names, command-line flags) is grepped against the references so accidental renames are caught; and --both confirms the .claude mirror is byte-identical. Keep those invariants when editing.

Reference map

Start at references/INDEX.md. Metadata: references/topics.json (topic → file + keywords).

Official documentation

Includes original content written for this skill. The official documentation is at https://docs.gimp.org/ — consult it to verify anything uncertain, conflicting, or version-specific.

What ships with it: 23 files

101.8 KB alongside SKILL.md

Gives 0 of the 12 instructions most mcp tooling skills give in ~1.3k tokens

Counted across 638 of the 750 authors here whose files we hold, read 2026-08-07

  • Create ten complex or independent read-only evaluation questionsin 69 of 638, across 15 files
  • Test servers using MCP Inspectorin 61 of 638, across 19 files
  • Provide actionable error messages with specific next stepsin 54 of 638, across 12 files
  • Prioritize comprehensive API coverage over specific workflows or workflow toolsin 54 of 638, across 12 files
  • Use TypeScript and Streamable HTTP for remote servers or clientsin 54 of 638, across 8 files
  • Define structured output schemas where possiblein 50 of 638, across 8 files
  • Use Zod or Pydantic for input schemasin 47 of 638, across 5 files
  • Fetch MCP specification pages with markdown suffixin 46 of 638, across 4 files
  • Load framework documentation using WebFetchin 45 of 638, across 3 files
  • Verify each evaluation answer independentlyin 45 of 638, across 3 files
  • Implement API client with authentication and paginationin 45 of 638, across 3 files
  • Define input schemas with validationin 27 of 638, across 9 files

Said here and by no other author read

  • consult the task router for relevant reference files
  • open only relevant reference files
  • answer strictly from reference files
  • mark uncovered information as TODO
  • verify procedure names before relying on them
  • use the in-app Procedure Browser for exact names

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.