agentleFS
HarrypatriaHarrypatria/tastemaker/skills/taste-brief/SKILL.md

taste-brief

Skill4 starsChanged 16 days agoMITView on GitHub

Turns a PRD, feature list, user story dump, screenshot or one-line app idea into a BUILD_MANIFEST.md (users, jobs, flows, views, states, data contracts, component inventory) before any UI is generated. Use when the user says "build this app", pastes requirements, or starts a new product in a vibe-coding tool. Part of the Tastemaker suite; prevents scope drift and generic layouts.

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:

    Use references/manifest-template.md.

    line 23
  2. Very specific:

    Never invent business facts (prices, customer names, compliance claims).

    line 27

≈ “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 Brief: requirements to Build Manifest

Generic UI comes from generic understanding. This skill extracts what is specific about the product, then fixes it in a manifest every later step reads.

Procedure

  1. Extract from the input: product one-liner, primary and secondary users, the 3 jobs each user hires the product for, risk level (regulated, money, health, safety), data density, devices, locale/RTL, existing stack and brand assets.
  2. Classify the archetype (drives taste-direction): data-workspace · crud-admin · consumer-app · marketing-site · content/editorial · commerce · devtool · form-heavy/workflow · realtime/monitoring.
  3. Derive flows: for each job, the shortest path as numbered steps ending in a success state. Mark the one critical flow.
  4. Derive views from flows, not from habit. Each view lists: purpose (one line), the user's top 3 questions on that view, primary action, data it reads/writes.
  5. State matrix per view: loading, empty, error, permission, success, plus domain states (e.g. "invoice overdue").
  6. Data contracts: METHOD /path -> TypeName with a TypeScript type per entity (realistic field names, units, enums).
  7. Component inventory bottom-up (atoms, molecules, organisms, templates), each marked reuse (exists in kit/project) or new. Delete anything not traceable to a flow.
  8. Gaps: list unknowns. If one would change architecture, ask (max 3 questions, one message). Otherwise record an assumption and continue.

Output: BUILD_MANIFEST.md

Use references/manifest-template.md. Keep it under 120 lines; it is a working document, not a spec novel. In chat platforms show a 10-line summary and keep the rest internal unless asked.

Rules

  • Never invent business facts (prices, customer names, compliance claims). Mark them ASSUMPTION: or DATA NEEDED:.
  • Seed data must look true for the domain: plausible names across cultures, realistic amounts and dates, edge cases (long names, zero values, nulls).
  • Screenshot or competitor input: extract structure and jobs, never copy brand, copy or distinctive visuals.
  • Archetype + top-3-questions are mandatory; taste-direction refuses to choose a layout without them.

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.