agentleFS
Sign inStart free
HarrypatriaHarrypatria/tastemaker/skills/taste-guard/SKILL.md

taste-guard

Skill4 starsChanged 16 days agoMITView on GitHub

Quality gate and fix loop of the Tastemaker suite. Runs the slop scanner, the state matrix, an accessibility trace and a compile trace on generated UI, then fixes findings until clean (max 3 cycles). Use after generating or editing any frontend code, before saying "done", when a hook reports slop findings, or when the user says "check", "verify", "lint the UI", "is this production ready".

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run after every BUILD step of the Taste Loop and on every hook report.

    line 8
  2. Very specific:

    No shell (chat platforms): apply the catalog's Auto rules by reading the code, top to bottom, and list findings in the same ID file:line …

    line 17

≈ “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

Taste Guard: nothing ships with slop

Run after every BUILD step of the Taste Loop and on every hook report. Findings use catalog IDs from tastemaker/references/slop-catalog.md.

Pass 1: Slop scan (automated)

node <skill-dir>/scripts/slop-scan.mjs <changed paths or src>      # human report
node <skill-dir>/scripts/slop-scan.mjs src --json                  # machine report
node <skill-dir>/scripts/slop-scan.mjs . --project                 # adds stack checks (E11-E13)

No shell (chat platforms): apply the catalog's Auto rules by reading the code, top to bottom, and list findings in the same ID file:line format.

Fix every error. Fix every warn or keep it with an taste-ignore <ID> <reason> comment the user can see. Info is optional polish.

Pass 2: State matrix (review)

For each view and each async organism, confirm it renders: loading (skeleton at final size), empty (teaches next action), filtered-empty, error (cause + retry, input kept), success, permission-denied; for each control: hover, focus-visible, active, disabled, loading. Missing state = U10 error. Show the matrix only if asked.

Pass 3: Accessibility trace (review)

Keyboard path through the critical flow (Tab order, Enter/Space, Esc returns focus); landmarks and heading order; names for every control; error association; contrast of any new color pair (taste-tokens/scripts/contrast.mjs); reduced motion; 320px reflow; color-only meaning (U15).

Pass 4: Compile trace (review, or run)

Imports resolve, props and types match, handlers wired, 'use client' only on interactive leaves, no unused code, keys stable. File-system environments: run the project's typecheck, lint and build and read the real output.

Fix loop

scan -> fix errors -> re-scan touched files
stop when: 0 errors on two consecutive passes, or 3 cycles done
after 3 cycles: report remaining findings honestly with IDs; never claim clean

Fix the cause, not the symptom: a V08 raw hex is fixed by adding or using a token, not by moving the hex into a variable inside the component.

Report line

Taste: <errors> errors · <warnings> warnings · score <0-100> · BASIC <n>/10 · build: passed|failed|not run

Claude Code hook

The plugin's PostToolUse hook runs the scanner on every edited UI file and returns error findings to you as feedback. Treat them as part of this loop. Control with TASTEMAKER_HOOK=off|warn|block (default block: errors returned as feedback; warn: printed only).

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.

Check your setup, free