agentleFS
wrg32786wrg32786/aigent-design-system/skills/design-resolver/SKILL.md

design-resolver

Skill4 starsChanged 3 months agoMITView on GitHub

Run the AIgent render, rank, repair, and verification loop on a website, deck, or product interface while preserving product truth, visual authority, inspiration constraints, accessibility, and performance.

The file

We split this skill into the rules it gives an agent

12 rules. About 0 are common, 9 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈9

    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. Very specific:

    Use this skill after the surface works end to end.

    line 8
  2. Very specific:

    Read .aigent/resolve/latest.md and .aigent/resolve/latest.json.

    line 37
  3. Distinctive:

    Keep product claims, selected visual world, reference transformations, responsive intent, and media strategy intact.

    line 40
  4. Distinctive:

    Run the smallest relevant code check.

    line 41
  5. Distinctive:

    When the mechanical gate passes, run Vision prepare and open every generated capture.

    line 43
  6. Distinctive:

    Do not flatten a distinctive composition merely to raise a mechanical score.

    line 48
  7. Distinctive:

    Do not remove meaningful motion when a reduced-motion alternative is the actual missing state.

    line 49
  8. Distinctive:

    Do not hide overflow until the overflowing element and layout contract are understood.

    line 50
  9. Distinctive:

    Do not add a dependency for a fix the current stack or browser already supports.

    line 51
  10. Distinctive:

    Do not change copy, assets, or visual identity to imitate an inspiration source.

    line 52
  11. Very specific:

    When the detector is wrong, record the deliberate exception in aigent.resolve.json or the final review rather than distorting the product.

    line 54
  12. Distinctive:

    no P0 or P1 issue remains.

    line 61

≈ “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

Design Resolver

Use this skill after the surface works end to end. It owns mechanical Resolve; route the required rendered judgment to visual-design-critic.

Read first

  1. PRODUCT.md
  2. DESIGN.md
  3. aigent.resolve.json
  4. .aigent/inspiration-plan.json, when present
  5. the current rendered surface

Initialize

npx github:wrg32786/aigent-design-system resolve --init --target .

For a running application, pass its exact local URL:

npx github:wrg32786/aigent-design-system resolve \
  --target . \
  --url http://127.0.0.1:3000/

Loop

Repeat no more than three coherent repair cycles before asking for human direction:

  1. Run Resolve.
  2. Read .aigent/resolve/latest.md and .aigent/resolve/latest.json.
  3. Inspect the top-ranked finding in the rendered page and trace the shared code path that owns it.
  4. Fix the root cause once. Do not patch sibling symptoms separately.
  5. Keep product claims, selected visual world, reference transformations, responsive intent, and media strategy intact.
  6. Run the smallest relevant code check.
  7. Rerun Resolve and compare resolved, introduced, and persistent findings.
  8. When the mechanical gate passes, run Vision prepare and open every generated capture.
  9. Complete Vision check and finalize before claiming completion.

Repair rules

  • Do not flatten a distinctive composition merely to raise a mechanical score.
  • Do not remove meaningful motion when a reduced-motion alternative is the actual missing state.
  • Do not hide overflow until the overflowing element and layout contract are understood.
  • Do not add a dependency for a fix the current stack or browser already supports.
  • Do not change copy, assets, or visual identity to imitate an inspiration source.
  • Resolve shared primitives before one-off instances.
  • When the detector is wrong, record the deliberate exception in aigent.resolve.json or the final review rather than distorting the product.

Stop condition

Stop only when:

  • the mechanical gate passes;
  • no P0 or P1 issue remains;
  • desktop, tablet, mobile, zoomed text, and reduced motion are complete;
  • product clarity and specificity remain intact;
  • composition, typography, motion, media, and originality receive an explicit visual review;
  • the structured Vision gate passes with no open P0/P1 finding;
  • the result has no unresolved loading, failure, rights, or provenance concern.

A green report without rendered judgment is not design completion.

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.