Skip to content

Art Direction Quality Loop

Rates visual work against a rubric using vision judges, iterates to a bar, and gates delivery.

Stars
19,682
Licence
MIT
Files
11 files

What it does

This skill adds a quality-control loop for artistic outputs such as images, films, slides, and site captures. Instead of accepting the first draft, the agent writes a brief, has a set of vision judges score the work against a written rubric, lists ranked defects to fix, and repeats up to four rounds. Only a passing result is handed over, and the user's own scores can be recorded to refine the judges over time. It does not create the artwork itself; another skill must do that.

Great for

  • Creators who want an objective second opinion on visuals before publishing
  • Teams that need a repeatable pass/fail check on artistic deliverables
  • Users who already have an image, film, or slide generation skill and want a quality gate

Skip it if

  • Producing the artifact itself, since a separate domain skill is required (the skill only judges and gates)
  • Text writing audits, which the skill explicitly excludes
  • Scores treated as absolute truth: judges are AI models and their verdicts can diverge from human taste, so the owner's own eye remains the final check

Try asking

  • “Critique this image against my brief and tell me the ranked defects”
  • “Run the quality loop on this film until it reaches the bar, then deliver it”
  • “Rate the visuals in this slide deck and list what to fix”
  • “Record my score of 8 and note for this render so the judges can be calibrated”
Step-by-step guide: how to use Art Direction Quality LoopInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 11 original files, the MIT licence and an attribution note.

    ArtDirection.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip ArtDirection.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip ArtDirection.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Complete Setup.md first, including a no-cost proof run, before relying on critique results
  • Write a Brief.json using References/BriefExample.json as a template so judges have a clear audience, register and bar
  • Treat a one-vendor panel result as weaker evidence and check the vendor_missing flag in the report

What's inside

  • Intents/Calibrate.md
  • Intents/Critique.md
  • Intents/Deliver.md
  • Intents/Direction.md
  • Intents/Loop.md
  • References/BriefExample.json
  • References/Rubric.md
  • Setup.md
  • SKILL.md
  • Tools/Calibrate.ts
  • Tools/Deliver.ts

Redistributed unmodified from danielmiessler/LifeOS under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.