agentleFS
ryanschueryanschue/automated-product-engineer/skills/figma-to-app-feature/SKILL.md

figma-to-app-feature

Skill0 starsChanged 27 days agoMITView on GitHub

Turn Figma designs, screenshots, exports, and product notes into a production-ready feature through artifact intake, product and architecture contracts, implementation, automated and manual QA, retrospective fix-back, durable documentation, deferred work, and release handoff.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 3 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Treat design artifacts as product requirements, not CSS paste targets.

    line 8
  2. Distinctive:

    Ask only for missing information that would materially change behavior, architecture, persistence, or release risk.

    line 24
  3. Very specific:

    Use design-system for reusable tokens, primitives, variants, metadata, and previews.

    line 28
  4. Very specific:

    Use architecture-building for responsibility, state, rendering, integration, and pull-request boundaries.

    line 29
  5. Very specific:

    Use frontend-api-db for durable state, authorization, endpoints, schema, and client-cache behavior.

    line 30
  6. Very specific:

    Use responsive-design and design-polish for fluid geometry and screenshot-level refinement.

    line 31
  7. Very specific:

    Use pr-workflow, code-review, and security-review for packaging and independent gates.

    line 32
  8. Very specific:

    Use manual-ui-qa for the final agent-controlled browser workout.

    line 33
  9. Distinctive:

    Create or update two distinct artifacts:

    line 37

≈ “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

Figma To App Feature

Treat design artifacts as product requirements, not CSS paste targets. This skill coordinates the full delivery loop; load the specialist skills only when their boundary is active.

Read references/delivery-lifecycle.md for the full staged workflow and exit evidence. Read references/artifact-and-state-contract.md when auditing a design handoff, resolving state precedence, or planning focused tests. Use the repository's design-export examples as concrete shape guidance when they are available, but preserve the target application's architecture and data contracts.

Adaptive intake

Inspect the accessible Figma source, target repository, existing design system, routes, data contracts, and tests before asking for information. Establish the target frames and states, user outcome, required behavior and copy, persistence and permissions, responsive and accessibility rules, and acceptance evidence.

Ask only when a missing product or design decision would materially change implementation. If the Figma source lacks a required frame, state, token mapping, or interaction rule, generate one focused copy-ready request for the designer with known context filled in and the exact artifact or decision needed. Do not offer a generic prompt catalog. If the design contract is sufficient, proceed.

Intake contract

  • Pin the request, exact design nodes or exports, accepted stories, repository, base SHA, existing branches and pull requests, and available environments.
  • Inventory component-level and page-level artifacts separately. Capture default, loading, empty, error, selected, open, hover/focus, permission, destructive, dense, and responsive states that apply.
  • Establish a source hierarchy for conflicting design, product notes, running behavior, and code. Record assumptions and superseded decisions.
  • Ask only for missing information that would materially change behavior, architecture, persistence, or release risk.

Coordinate by boundary

  • Use design-system for reusable tokens, primitives, variants, metadata, and previews.
  • Use architecture-building for responsibility, state, rendering, integration, and pull-request boundaries.
  • Use frontend-api-db for durable state, authorization, endpoints, schema, and client-cache behavior.
  • Use responsive-design and design-polish for fluid geometry and screenshot-level refinement.
  • Use pr-workflow, code-review, and security-review for packaging and independent gates.
  • Use manual-ui-qa for the final agent-controlled browser workout. Playwright coverage is required where appropriate but does not satisfy this manual gate.

Durable knowledge

Create or update two distinct artifacts:

  • A living feature guide describing current purpose, stories, visible states, code map, data flow, state ownership, APIs, persistence, authorization, failure behavior, responsive/accessibility behavior, validation, and known risks.
  • A historical retrospective describing what happened during implementation and QA, what failed, what was learned about both the feature and delivery pipeline, and the disposition of every finding.

Promote reusable pipeline lessons into shared instructions, runbooks, tests, or skills. Record important non-blocking work in the authorized backlog with evidence, impact, safe-to-defer reasoning, a revisit trigger, and acceptance criteria.

Completion contract

Completion requires accepted stories on one exact SHA; appropriate focused, full, visual, accessibility, and Playwright checks; a distinct authenticated manual browser QA pass; resolved release blockers; retrospective fix-back of the highest-impact in-scope issues; current feature and pipeline documentation; durable deferred work; and final evidence.

When external communication is authorized, send a concise authorized issue tracker or team channel handoff containing the user outcome, exact SHA and environment, pull requests, test or preview link, automated and manual QA result, retrospective fixes, deferred tickets, and links to a useful screenshot and short video. Verify delivery. Never merge, release, change secrets, perform destructive data work, or send externally without the required authority.

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.