frontend-task
Skill8 starsChanged 3 months agoApache-2.0View on GitHub
Orchestrates the frontend agent team to plan, review, implement, and review frontend changes
In this repository7 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- add-shadcn-componentSkill
- frontend-taskSkill
- workspace-observer-onboardingSkill
- add-shadcn-componentSkill
- maskin-designSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Read the user's request and classify it:
- Very specific: line 21
Create a team with these three agents (defined in .claude/agents/):
- Distinctive: line 37
If the reviewer sends feedback, the planner addresses it and the reviewer re-checks
- Distinctive: line 49
If the reviewer sends feedback, the implementer fixes and the reviewer re-checks
- Distinctive: line 54
When all tasks are complete:
- Very specific: line 55
Run pnpm lint and pnpm type-check as a final safety net
- Very specific: line 64
If the task involves adding a new shadcn/ui component, use: cd apps/web && npx shadcn@latest add <component> --yes
≈ “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.
Files
Frontend Task Orchestration
You are orchestrating a frontend task using a team of specialized agents. Follow this workflow exactly.
Workflow
1. Assess the Task
Read the user's request and classify it:
- Trivial: typo fix, single-line change, obvious one-file fix → skip to Phase 2
- Non-trivial: new feature, multi-file change, UI addition, behavior change → start at Phase 1
2. Create the Team
Create a team with these three agents (defined in .claude/agents/):
frontend-planner— explores codebase and creates implementation planfrontend-reviewer— reviews plans and implementations for qualityfrontend-implementer— writes the code following approved plans
3. Phase 1: Plan + Review (skip for trivial tasks)
Create these tasks:
Task A: "Plan: {description of the frontend task}"
- Assign to:
frontend-planner - The planner will explore existing components, hooks, and patterns, then produce a structured plan
Task B: "Review the plan" (depends on Task A)
- Assign to:
frontend-reviewer - The reviewer checks the plan for component reuse, placement consistency, and rule compliance
- If the reviewer sends feedback, the planner addresses it and the reviewer re-checks
- Max 2 feedback rounds — after that, the reviewer approves with notes and we proceed
4. Phase 2: Implement + Review
Task C: "Implement the approved plan" (depends on Task B, or no dependency for trivial tasks)
- Assign to:
frontend-implementer - The implementer follows the approved plan exactly, then runs lint + type-check
Task D: "Review the implementation" (depends on Task C)
- Assign to:
frontend-reviewer - The reviewer checks the code against the rules and the approved plan
- If the reviewer sends feedback, the implementer fixes and the reviewer re-checks
- Max 2 feedback rounds — after that, the reviewer approves with notes
5. Finalize
When all tasks are complete:
- Run
pnpm lintandpnpm type-checkas a final safety net - Summarize what was done for the user
- List any "approved with notes" items that should be addressed later
Important Notes
- The agents communicate directly with each other for feedback loops — you don't need to relay messages
- Each agent reads
.claude/rules/frontend.mdandapps/web/CLAUDE.mdat startup - The reviewer is the quality gate — nothing ships without reviewer approval
- If the task involves adding a new shadcn/ui component, use:
cd apps/web && npx shadcn@latest add <component> --yes - For tasks that touch both frontend and backend, only handle the frontend portion here
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
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.

