agentleFS
ytvee-devytvee-dev/webdev-agent-kit/skills/greenfield-project-builder/SKILL.md

greenfield-project-builder

Skill7 starsChanged 21 days agoApache-2.0View on GitHub

Plan a new frontend project or first vertical slice from a product idea, with approval gates before scaffolding. Excludes existing-project micro-fixes, automatic scaffolds, package installation, UI libraries, and test setup.

The file

Show its rules and text

We split this skill into the rules it gives an agent

25 rules. About 0 are common, 16 are distinctive, and 9 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈16

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this skill only for Deep Workflow tasks.

    line 44
  2. Distinctive:

    Use this skill when the user asks to:

    line 46
  3. Distinctive:

    start a new frontend project.

    line 48
  4. Distinctive:

    create a project structure before scaffolding.

    line 50
  5. Distinctive:

    Do not use this skill for:

    line 58
  6. Very specific:

    Do not use this skill when frontend-architecture-planner is enough for an existing project.

    line 70
  7. Very specific:

    Read the compact goal contract or project/active-goals.md when present.

    line 79
  8. Very specific:

    Read the compact execution plan or project/active-plan.md when present.

    line 80
  9. Distinctive:

    Read existing project overlays only when this is a new app inside a larger existing workspace:

    line 81
  10. Distinctive:

    Must not install packages without explicit approval.

    line 95
  11. Distinctive:

    Must not create app source files, framework configs, package manifests, routes, components, styles, tests, or build scripts without …

    line 96
  12. Distinctive:

    Must not introduce UI component libraries or testing workflows.

    line 97
  13. Distinctive:

    Must not access production systems, secrets, or production data.

    line 98
  14. Distinctive:

    Keep backend, database, auth, payments, analytics, UI libraries, tests, and infrastructure out of scope unless explicitly requested and …

    line 126
  15. Very specific:

    Create a minimal architecture plan.

    line 128
  16. Very specific:

    Use frontend-architecture-planner when route/state/data/styling/build boundaries require deeper planning.

    line 129
  17. Very specific:

    Use project-onboarding-adapter to create only the host-root pointer and local-only project/ overlays until the user approves actual project …

    line 145
  18. Very specific:

    Use execution-plan-manager for slices after the greenfield plan is accepted.

    line 148
  19. Distinctive:

    Stop before scaffold unless approved.

    line 150
  20. Distinctive:

    If approval is missing, return a plan and exact approval request instead of creating files.

    line 151
  21. Distinctive:

    no scaffold was run without explicit approval.

    line 185
  22. Distinctive:

    no package was installed without explicit approval.

    line 186
  23. Distinctive:

    no UI component library was introduced.

    line 187
  24. Distinctive:

    no testing workflow was introduced.

    line 188
  25. Very specific:

    no application source/config/package/build files were created without explicit approval.

    line 189

≈ “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

Greenfield Project Builder

Purpose

Plan and execute the first safe steps of a new frontend project without turning a vague product idea into uncontrolled scaffolding.

This skill converts a product idea into a minimal frontend project plan, explicit stack assumptions, approval gates, first vertical slice, and onboarding handoff. It must not scaffold, install packages, or create application files unless the user explicitly approves the exact action.

When To Use

Use this skill only for Deep Workflow tasks.

Use this skill when the user asks to:

  • start a new frontend project;
  • plan a new frontend app from a product idea;
  • create a project structure before scaffolding;
  • define the first vertical slice of a new app;
  • choose an intended frontend stack before project creation;
  • prepare onboarding overlays for an empty or new project;
  • turn a broad product concept into a small frontend implementation path.

When Not To Use

Do not use this skill for:

  • existing-project bugfixes;
  • one route or component in an existing project;
  • purely visual redesign of an existing screen;
  • refactoring existing code;
  • UI component library setup;
  • testing workflows;
  • backend, database, ORM, migrations, infrastructure, or production systems;
  • package installation by default;
  • automatic scaffolding without approval.

Do not use this skill when frontend-architecture-planner is enough for an existing project.

Required Context

  1. Read AGENTS.md.
  2. Read common/prompt-intent-routing-rules.md.
  3. Confirm the task is Deep Workflow.
  4. Read common/framework-adaptation-policy.md and common/build-tool-boundary-rules.md.
  5. Read the compact goal contract or project/active-goals.md when present.
  6. Read the compact execution plan or project/active-plan.md when present.
  7. Read existing project overlays only when this is a new app inside a larger existing workspace:
    • project/stack-profile.md;
    • project/architecture-map.md;
    • project/build-profile.md when present;
    • project/workspace-profile.md when present;
    • project/verification-profile.md when present.

Tool Contract

  • May inspect the repository root and selected workspace files to determine whether the target is empty, new, or inside an existing workspace.
  • Activate openai_platform_docs only when the planned product uses OpenAI APIs, ChatGPT Apps SDK, or Codex integration behavior.
  • May write local-only project/** overlays after approval or during approved onboarding.
  • May create a host-root AGENTS.md pointer only through project-onboarding-adapter rules.
  • May propose scaffold commands, but must not run them without explicit approval.
  • Must not install packages without explicit approval.
  • Must not create app source files, framework configs, package manifests, routes, components, styles, tests, or build scripts without explicit approval.
  • Must not introduce UI component libraries or testing workflows.
  • Must not access production systems, secrets, or production data.

Workflow

  1. Confirm greenfield status.
    • New standalone project.
    • Empty repository.
    • New app inside existing workspace.
    • Existing project that actually needs architecture planning instead.
  2. Define the product goal. Route to goal-planner first if the product goal is unclear.
  3. Define the first vertical slice. The first slice should include one user-visible path through the product, not a full architecture.
  4. Define intended stack assumptions. Include only what is known or explicitly approved:
    • framework;
    • language;
    • routing model;
    • styling approach;
    • package manager;
    • build tool;
    • deployment target if relevant;
    • verification command expectations.
  5. Define non-goals. Keep backend, database, auth, payments, analytics, UI libraries, tests, and infrastructure out of scope unless explicitly requested and approved.
  6. Create a minimal architecture plan. Use frontend-architecture-planner when route/state/data/styling/build boundaries require deeper planning.
  7. Define approval gates. Require explicit approval for:
    • scaffold commands;
    • package installation;
    • package manager choice or change;
    • framework choice or change;
    • UI component library;
    • state/data/form/styling libraries;
    • testing setup;
    • build tooling;
    • repository structure changes;
    • destructive file operations.
  8. Create onboarding handoff. Use project-onboarding-adapter to create only the host-root pointer and local-only project/** overlays until the user approves actual project scaffolding.
  9. Create execution plan. Use execution-plan-manager for slices after the greenfield plan is accepted.
  10. Stop before scaffold unless approved. If approval is missing, return a plan and exact approval request instead of creating files.

Output Contract

Final response: return only facts that affect the user's understanding, confidence, or next action. Omit empty fields and workflow narration.

Return or write a Greenfield Project Plan with:

Product Goal
Workflow Level
Greenfield Status
First Vertical Slice
Intended Stack Assumptions
Non-Goals
Minimal Architecture
Files Or Surfaces To Create After Approval
Approval Gates
Onboarding Overlay Plan
Implementation Slices
Verification Strategy
Risks
Open Questions
Next Approval Request Or Next Step

Omit irrelevant sections rather than filling them with generic text.

Validation Gates

Before finishing, verify:

  • the task is genuinely greenfield or empty-project work;
  • the first vertical slice is smaller than the whole product;
  • no scaffold was run without explicit approval;
  • no package was installed without explicit approval;
  • no UI component library was introduced;
  • no testing workflow was introduced;
  • no application source/config/package/build files were created without explicit approval;
  • host-project facts stay in project/**;
  • next approval request is precise if action is blocked.

Trigger Evals

Should trigger:

  • "Start a new frontend project for this product idea."
  • "Plan the first version of a new Next app, but do not scaffold yet."
  • "Create a greenfield frontend plan and tell me what you need approval for."
  • "This repo is empty; prepare the frontend project structure safely."
  • "Define the first vertical slice before creating the app."

Should not trigger:

  • "Fix this TypeScript error."
  • "Add a card to this existing dashboard."
  • "Make this existing page look better."
  • "Plan the architecture of this existing module."
  • "Install shadcn."
  • "Create tests."

Reference Map

  • AGENTS.md - canonical policy, routing, tool rules, and documentation rules.
  • common/prompt-intent-routing-rules.md - workflow weight selection and escalation rules.
  • common/framework-adaptation-policy.md - intended stack adaptation rules.
  • common/build-tool-boundary-rules.md - build and workspace approval gates.
  • common/frontend-integration-boundaries.md - frontend integration scope limits.
  • skills/goal-planner/SKILL.md - product goal contract.
  • skills/execution-plan-manager/SKILL.md - task slicing and stop/resume planning.
  • skills/frontend-architecture-planner/SKILL.md - architecture boundary planning.
  • skills/project-onboarding-adapter/SKILL.md - safe host-root pointer and local-only overlay creation.
  • project/stack-profile.md - local-only intended or detected stack facts.
  • project/architecture-map.md - local-only intended or detected architecture facts.
  • project/verification-profile.md - local-only verification command facts.

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.