agentleFS
syi0808syi0808/svg-skills/illustration-plan/SKILL.md

illustration-plan

Skill4 starsChanged 3 months agoMITView on GitHub

Plan the visual direction and semantic parts of an illustration before production. Use for mood and style exploration, art-direction questions, composition planning, or preparing a scene.json for vector or raster artwork; not SVG authoring, image generation, or faithful image-to-SVG conversion by itself.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 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.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the brief and any supplied plan or reference.

    line 14
  2. Distinctive:

    Ask only about missing choices that materially change the result.

    line 31
  3. Distinctive:

    Do not promise photographic fidelity or unsupported effects.

    line 38
  4. Very specific:

    When a direction is chosen or production needs a handoff, write scene.json using scene-model.md.

    line 44
  5. Very specific:

    Use version 2 with phase: "plan", a concise art_direction.description, a canvas, and meaningful element IDs and roles.

    line 46
  6. Very specific:

    Do not precompute path commands or control points.

    line 50
  7. Distinctive:

    Do not demote a bound map to a plan or discard implementation details accidentally.

    line 58

≈ “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

Illustration Plan

Turn the brief into a useful visual direction and semantic scene plan. Stop at planning: do not generate images, author SVG paths, or start production unless the user requested it. This skill works independently of the production skills.

Establish what is already decided

Read the brief and any supplied plan or reference. Inspect actual reference images when available; distinguish inspiration from an image that must be copied faithfully. For faithful conversion, extract the source's direction rather than inventing a new one. Preserve existing IDs, approved decisions, and source files.

Record the subject, intended use and display size, aspect ratio, background, required text or branding, and features that must remain unchanged. Do not turn a local edit into a new art-direction exercise. Update only affected decisions when continuing an existing plan.

Resolve visual direction

Separate mood, rendering language, and material rather than treating style names as mutually exclusive presets. A calm illustration can also be softly three- dimensional and glass-like. Read art-direction.md when the brief needs style exploration or feasibility decisions.

Ask only about missing choices that materially change the result. Use plain language or a few distinct directions when the user does not know design terms; do not require a questionnaire. Infer low-impact defaults from context and make important assumptions visible. If the user asks only for options, return those options and stop rather than silently selecting one and producing artwork.

For SVG output, balance appearance with editability and the intended renderer. Do not promise photographic fidelity or unsupported effects. The current bundled SVG tools support gradients and opacity but not SVG filters, including blur. Explain a consequential approximation or renderer requirement before adopting it.

Write the semantic plan

When a direction is chosen or production needs a handoff, write scene.json using scene-model.md. Keep simple plans short. Use version 2 with phase: "plan", a concise art_direction.description, a canvas, and meaningful element IDs and roles. Record only useful assumptions, palette hints, and spatial or preservation constraints.

Do not precompute path commands or control points. Leave actual SVG tag and parent fields unset: the author adds these and changes the phase to bound after constructing and inspecting the SVG. Planned relationships belong in plain-language constraints. A plan is not a geometry compiler or executable prompt.

Existing version 1 scenes remain supported bound maps; do not migrate them just to make a local edit. When deliberately adopting v2 art direction for an existing SVG, retain its mapping as bound and have the production workflow verify it. Do not demote a bound map to a plan or discard implementation details accidentally.

Validate and hand off

Resolve SKILL_DIR to this skill's directory. Python 3.10+ is sufficient:

python3 "$SKILL_DIR/scripts/validate_scene.py" scene.json

This checks the scene contract without needing an SVG or renderer. It does not judge design quality or verify actual SVG mappings. Report unresolved assumptions and summarize the selected direction in the user's language. Return the plan path when one was written; do not automatically call a production skill. When another workflow requested planning, hand back the plan and let that workflow continue.

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.