impl-ui

Implements UI features by orchestrating disposable worker agents from a written implementation brief.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/muselesscreator/ui-skills --skill impl-ui-muselesscreator
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: impl-ui
Source: https://github.com/muselesscreator/ui-skills/tree/main/impl-ui
Command: npx skills add https://github.com/muselesscreator/ui-skills --skill impl-ui-muselesscreator

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Implementing a UI feature directly in an interactive session bloats the context window with file reads and edits, leaving little room for the rest of the conversation. This Skill keeps the calling session lean by acting as a thin orchestrator: it writes a compact implementation brief to disk and spawns fresh, disposable worker agents that do the actual file reading and code writing. ## Core Features & Use Cases - Plan-driven implementation: Picks up the latest /plan-ui output (or a task description), honors resolved decision memos, and refuses to start while decisions are still open. - Worker orchestration: Spawns one builder agent per logical file group, a sonnet presentation-layer pass for semantic markup and styling, and a read-only standards/test-gap check. - Scoped safety gates: Checks for collisions only against the plan's own files, runs a scoped TypeScript type gate before validation, and resolves worker-raised decision memos inline via AskUserQuestion. - Use Case: After planning a new settings panel with /plan-ui, run this Skill to have workers implement the components, refine accessibility semantics, verify types compile, and receive a report listing the unit and E2E tests still needed. ## Quick Start Ask the AI to implement the planned UI feature using the most recent plan, for example by saying "implement what we planned with impl-ui".

Frequently Asked Questions about impl-ui

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I implement a planned UI feature without bloating my context window?

Use an orchestrator approach: write a compact implementation brief to disk, then spawn fresh worker agents that read only the files their group owns. The calling session passes file paths instead of contents, adding roughly 10k tokens while each disposable worker uses its own context.

How do I run impl-ui after creating a plan?

Pass "use plan" as the argument and the Skill locates the most recent plan file automatically. It refuses to start if any decision memos from the plan are still open, surfacing their ids so you can resolve them first.

Does impl-ui check the whole git working tree for conflicts?

No, collision checks are scoped to the files listed in the plan via a dedicated script. A dirty tree that does not intersect the plan's files is treated as unrelated parallel work and does not block implementation.

What happens when a worker hits an unresolved ambiguity during implementation?

The worker writes a decision memo with lettered options and continues with its unblocked files. The orchestrator then presents each memo via AskUserQuestion, records the resolution, and respawns only the workers that had blocked files.

Does impl-ui run linting and tests after writing code?

No, it runs only a scoped TypeScript type gate to catch compile errors before behavioral validation. Lint, Prettier, dead code, and test fixes are deliberately deferred to the separate cleanup-ui skill.