Execute qualquer Skill no Manus
com um clique

Execute qualquer Skill no Manus com um clique

Começar

existing-project

Estrelas8

Forks0

Atualizado4 de maio de 2026 às 11:51

Use when adding project structure and rules to an existing codebase that has no CLAUDE.md or AGENTS.md

Instalação

Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.

Executar no Manus

Fonte

sheikhlimon

sheikhlimon/dotfiles

Abrir repositório GitHub Ver repositórios do creator

Download

Executar no Manus

Ocupações relacionadasSOC

Baseado na classificação ocupacional SOC

Desenvolvedores de softwareInformática e Matemática·SOC 15-1252

SKILL.md

readonly

name	existing-project
description	Use when adding project structure and rules to an existing codebase that has no CLAUDE.md or AGENTS.md

Existing Project Setup

Audits an existing codebase and scaffolds CLAUDE.md, AGENTS.md, and docs/plan.md.

Process

Step 1: Audit

Read the codebase before writing anything:

What frameworks, libraries, tools are in use (check package.json, config files)
What commands exist (check scripts in package.json)
What file organization exists (ls the directory tree)
What testing is set up (test config, existing test files)
What patterns are already used (naming, state management, styling)
What's broken or missing (failing tests, TODO comments, missing configs)

Step 2: Ask

Confirm with the user:

Is the audit accurate? Anything I missed?
Any working rules they want enforced?
Any design preferences?

Step 3: Create Files

CLAUDE.md (project root, auto-read by Claude Code)

Based on audit, not guessing. Keep under 80 lines.

Must include:

What the project does (from audit)
Current state (from audit — what works, what doesn't)
Tech stack and commands (from package.json)
@AGENTS.md — import for all coding conventions
Before starting any session: read plan, follow AGENTS.md rules

AGENTS.md (project root, imported by CLAUDE.md)

Conventions based on patterns found in the codebase. Keep under 100 lines.

Must include:

Working rules (explain first, one logical unit per response, commit after each feature, answer questions, DRY)
Code style (from existing patterns — don't invent new conventions)
Testing approach (from what's already set up)
File organization (map what actually exists)
Git conventions (from commit history patterns)
Anti-patterns and preferences

docs/plan.md

What's broken (from audit)
What's missing (from audit)
Phased plan with checkboxes

Rules

Audit first, write second. Never guess.
CLAUDE.md = project context + imports. AGENTS.md = coding rules. No overlap.
Respect existing conventions — don't force new patterns on an established codebase
No fluff

Mais deste repositório

mesmo repositório

mentor-mode

sheikhlimon/dotfiles

Use when explaining unfamiliar concepts, architectures, bugs, or implementation patterns. Focus on mental models, why the pattern exists, and transfer to other contexts instead of surface-level code explanation.

2026-07-108

collaborator-tone

sheikhlimon/dotfiles

Write responses in a thoughtful, technically confident, collaborative engineering tone suitable for GitHub, Discord, OSS discussions, reviews, chats, and maintainer conversations.

2026-05-168

explain-code

sheikhlimon/dotfiles

Use when the user wants to understand existing code — what it does, why it exists, and how to recall it later. Not for teaching during implementation (use mentor-mode). Use when the user asks "what does this file do", "explain this code", or points at code they don't understand.

2026-05-148

just-rules

sheikhlimon/dotfiles

Use when wanting minimal project rules without design docs or implementation plans — just CLAUDE.md and AGENTS.md

2026-05-048

new-project

sheikhlimon/dotfiles

Use when starting a brand new project from scratch and needing project structure, rules, and documentation files generated

2026-05-048

name	existing-project
description	Use when adding project structure and rules to an existing codebase that has no CLAUDE.md or AGENTS.md

Existing Project Setup

Audits an existing codebase and scaffolds CLAUDE.md, AGENTS.md, and docs/plan.md.

Process

Step 1: Audit

Read the codebase before writing anything:

What frameworks, libraries, tools are in use (check package.json, config files)
What commands exist (check scripts in package.json)
What file organization exists (ls the directory tree)
What testing is set up (test config, existing test files)
What patterns are already used (naming, state management, styling)
What's broken or missing (failing tests, TODO comments, missing configs)

Step 2: Ask

Confirm with the user:

Is the audit accurate? Anything I missed?
Any working rules they want enforced?
Any design preferences?

Step 3: Create Files

CLAUDE.md (project root, auto-read by Claude Code)

Based on audit, not guessing. Keep under 80 lines.

Must include:

What the project does (from audit)
Current state (from audit — what works, what doesn't)
Tech stack and commands (from package.json)
@AGENTS.md — import for all coding conventions
Before starting any session: read plan, follow AGENTS.md rules

AGENTS.md (project root, imported by CLAUDE.md)

Conventions based on patterns found in the codebase. Keep under 100 lines.

Must include:

Working rules (explain first, one logical unit per response, commit after each feature, answer questions, DRY)
Code style (from existing patterns — don't invent new conventions)
Testing approach (from what's already set up)
File organization (map what actually exists)
Git conventions (from commit history patterns)
Anti-patterns and preferences

docs/plan.md

What's broken (from audit)
What's missing (from audit)
Phased plan with checkboxes

Rules

Audit first, write second. Never guess.
CLAUDE.md = project context + imports. AGENTS.md = coding rules. No overlap.
Respect existing conventions — don't force new patterns on an established codebase
No fluff