agentleFS
SamuelAsherRivelloSamuelAsherRivello/ai-skills-blender/.claude/skills/blender-create-visual-target-2d/SKILL.md

blender-create-visual-target-2d

Skill11 starsChanged 36 days agoMITView on GitHub

Generate or interpret a prompt-specific 2D visual target and 3D direction brief. Start as a parallel target worker for every new Blender 3D task and use its result for preview feedback; also callable directly for concept direction.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are 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.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the subject, intended 3D use and available images.

    line 22
  2. Distinctive:

    Write a concise 3D direction brief: primary and secondary forms, relative proportions, material families, light/camera cues, prioritized …

    line 29

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Blender Create Visual Target 2D

Use the existing MCP connection whenever it can perform the operation. On Windows, auxiliary execution MUST remain windowless, including brief flashes, during setup, diagnostics, rendering, export, and verification. Before launching, establish suppression for the complete process chain, including client-owned and dependency-owned startup. Use explicit no-console creation (such as subprocess.CREATE_NO_WINDOW) and captured output for reviewed console-only commands. If any required boundary is unknown or unsupported, do not launch: report BLOCKED with the reason and next action, and continue independent safe work. Never retry through a visible terminal, toggle Blender's system console, or weaken suppression. An absolute executable path, cmd /c, Blender --background, or Start-Process -WindowStyle Hidden alone is not proof of no flashes. Keep exit status, bounded timeouts, useful diagnostics without secrets, and cleanup of owned descendants. Do not hide, minimize, restore, focus, or close the existing Blender editor to suppress auxiliary windows. Preserve fresh editor-readiness checks before screenshots.

Inputs and scope

Accept a subject or supplied target image, intended 3D use, and optional style, setting, pose/action, camera, lighting, palette, aspect ratio, resolution, references and budget. State reasonable defaults; clarify only material ambiguity. Every new 3D task starts a prompt-specific visual-target worker by default, alongside the main Blender work. A target-only request does not require Blender, its MCP connection, or any scene mutation.

For 3D-task orchestration, read references/parallel-feedback.md. The parent launches one worker and continues construction immediately; nested skills reuse the same task/target. When running as the target worker, execute the workflow below without spawning another worker or changing Blender. A direct target-only invocation executes it once, with no recursive delegation.

Use an image tool actually available in the current session and follow its provider instructions. Codex or Claude being the client does not establish image-generation capability. Do not install a provider, switch to an unconfigured API, acquire credentials or start a paid service as a fallback. A supplied image can be analyzed without generation; otherwise deliver an explicit prompt-only result when generation is unavailable.

Targets are concept inputs for inspiration and comparison. Preserve genuine Blender renders, editor screenshots, meshes and textures. This skill does not reconstruct geometry, make billboards or project the image onto a model; blender-convert-2d-3d handles separately requested reconstruction.

Workflow

  1. Read the subject, intended 3D use and available images. Select new generation, supplied-target interpretation, or an explicitly requested revision.
  2. Establish style, setting, pose, framing, lighting, palette and output constraints. When camera is unspecified, choose a clear diagnostic view that exposes the subject's important forms; preserve intentional stylization.
  3. Inspect any references. Separate required content from inspiration and identify conflicts with explicit user choices; the image does not silently override the requested subject, lighting or layout.
  4. Discover an available authorized image tool and relevant usage instructions. If none exists, use a supplied target or deliver the prompt-only path with missing-image status. Do not invent a successful tool call.
  5. Compose and save the exact generation prompt, including essential forms, requested style and complete subject framing. For supplied imagery, identify the current interpretation request and mark the original generation prompt unknown unless provided.
  6. Generate one target by default, or preserve the supplied image. Retain previous revisions and follow the requested variant budget. Save a project-bound image inside the project; record actual size/format rather than silently cropping or claiming unsupported dimensions.
  7. Inspect the actual image for complete subjects, pose, style, camera, material cues and contradictory or implausible details. Record essential mismatches as needs-revision. Do not enter an unbounded generation loop or equate aesthetic polish with geometric accuracy.
  8. Write a concise 3D direction brief: primary and secondary forms, relative proportions, material families, light/camera cues, prioritized visible details, inferred hidden geometry and budget-sensitive simplifications. Distinguish observations from assumptions and avoid promising photographic parity.
  9. Save a versioned target package following references/target-package.md. Deliver its image, prompt, brief and manifest with explicit readiness status. As a worker, return to the parent with prioritized cues and limitations; do not enter Blender creation skills yourself. A direct target-only request ends with this handoff.
  10. The parent folds the completed target into the current 3D task: inspect it beside a genuine preview, rank visible gaps, make authorized corrections and compare again using the same target version. Follow the bounded feedback loop in parallel-feedback.md. Without a genuine preview, return criteria without claiming a 3D evaluation.

Handoff

Use blender-create-model for props/characters and blender-create-environment for scenes. A target defines priorities, not a measurement survey: keep dimensions and unseen geometry editable. During a build/review cycle, freeze the chosen target; disclose a new target version as a changed comparison baseline. Match pose/view where useful, but assess style and structure rather than demand pixel identity.

The target may contain extra signage, objects or visual errors. Carry forward only cues consistent with the request. Keep generated concept imagery out of result.png and editor.png destinations reserved for real Blender output. A comparative report can show both with explicit labels and original links.

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.