analyze-task

Gathers repo learnings, branch state, and tooling constraints into a task-context artifact.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Before planning UI work, developers waste time re-discovering repo conventions, in-flight branch changes, and tooling rules that will block implementation. This Skill collects those fixed facts once into a single bounded artifact that downstream planning, implementation, and review skills can consume. ## Core Features & Use Cases - Fact gathering without planning: Collects repo learnings, git branch state, TypeScript/ESLint/Prettier constraints, and user-named files while deliberately leaving affected-file discovery to the planning stage. - Handoff and decision seeding: Reuses prior savepoint handoffs and resolved decision memos so settled questions are not re-litigated. - Bounded artifact output: Writes a distilled analysis file (target under 150 lines) with a fixed frontmatter schema, symlinked as analysis-latest.md for /plan-ui, /impl-ui, /validate-ui, and /pr-review-ui. - Use Case: Before asking for a plan to add a new settings form, run this Skill to capture which files the task names, what in-flight branch edits might conflict, and which lint rules will reject common patterns. ## Quick Start Ask the assistant to analyze this task and gather the constraints for adding a new settings form before any planning begins.

Frequently Asked Questions about analyze-task

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

FAQPage Schema
How do I gather context before planning a UI task?

Run the analyze-task skill with a description of the UI task. It collects repo learnings, branch state, tooling rules, and any files the task names into a single analysis artifact that planning skills read directly.

What is the difference between mentioned files and affected files?

Mentioned files are ones the user explicitly names in the task and are read during analysis as upfront metadata. Affected files are what the change turns out to touch, which is a planning conclusion deliberately left to the planning step.

Does analyze-task work without prior repo learnings?

Yes, but it warns that no learnings were found and proceeds with live analysis only, which is slower. For a work repo run /wiki-index first; for an OSS checkout run /learn-repo to build the learnings it reads.

Can analyze-task reuse a previous savepoint handoff?

Yes. If a handoff-latest.md exists, its key files, decisions, and gotchas seed the artifact with provenance noted. Live git state remains the source of truth when the handoff and the branch disagree.

What happens if a step in the analysis fails?

The artifact is still written with an ERROR section at the top describing what failed, such as unreachable learnings or unreadable config. The skill never fails silently, so downstream steps know the gap exists.