Project Onboarding

Analyze the current project and save comprehensive context as Cortex memories.

Arguments

$ARGUMENTS

Flags

• --refresh: Update existing memories instead of creating new ones
• --only <type>: Analyze only specific aspect (stack, architecture, database, testing)
• --dry-run: Show what would be saved without actually saving

Your Task

Analyze this project and generate memories for each aspect below. Use the tools available to explore the codebase.

Step 1: Detect Project Type

Check for these files to identify the stack:

File	Stack
go.mod	Go
mix.exs	Elixir
package.json	Node.js
requirements.txt / pyproject.toml	Python
Cargo.toml	Rust
Gemfile	Ruby

Step 2: Analyze Each Aspect

2.1 Stack Analysis (implementation:project:stack)

Gather:

• Language and version
• Framework (if any)
• Key dependencies (top 10-15 most important)
• Package manager
• Build commands

Save as:

mcp__cortex__memory(action="save", 
  type="implementation",
  title="project:stack",
  content="..."
)

2.2 Architecture Analysis (architecture:project:structure)

Gather:

• Directory structure (top 2-3 levels)
• Architectural pattern (MVC, DDD, Clean, Contexts, etc.)
• Entry points (main files, routers)
• Key modules/packages

Save as:

mcp__cortex__memory(action="save", 
  type="architecture",
  title="project:structure",
  content="..."
)

2.3 Database Analysis (implementation:project:database)

If database exists, gather:

• Database type (Postgres, MySQL, SQLite, MongoDB)
• ORM/Query builder
• Schema/Models location
• Key tables/collections with relationships
• Important indexes
• Migration commands

Detection by stack:

• Ecto: priv/repo/migrations/, lib/**/schemas/
• GORM: structs with gorm: tags
• Prisma: prisma/schema.prisma
• TypeORM: **/entities/*.ts
• Rails: db/schema.rb
• Django: */models.py
• SQLite raw: check for .db files

Save as:

mcp__cortex__memory(action="save", 
  type="implementation",
  title="project:database",
  content="..."
)

2.4 Testing Analysis (testing:project:testing)

Gather:

• Test framework
• Test file locations
• How to run tests
• Coverage configuration (if any)
• CI/CD setup (GitHub Actions, etc.)

Save as:

mcp__cortex__memory(action="save", 
  type="testing",
  title="project:testing",
  content="..."
)

2.5 Project Overview (context:project:overview)

Create a summary with:

• Project name and purpose
• Main stack (1-2 lines)
• Key features/capabilities
• Quick start commands

Save as:

mcp__cortex__memory(action="save", 
  type="context",
  title="project:overview",
  content="..."
)

Step 3: Memory Content Templates

Stack Template

# Stack: [Language] [Version]

## Framework
[Framework name and version]

## Package Manager
[npm/go mod/mix/pip/cargo]

## Key Dependencies
- [dep1]: [purpose]
- [dep2]: [purpose]
...

## Build Commands
- Build: `[command]`
- Run: `[command]`
- Install deps: `[command]`

Architecture Template

# Architecture: [Pattern Name]

## Directory Structure

[tree output, 2-3 levels]

## Layers/Modules
- [layer1]: [purpose]
- [layer2]: [purpose]

## Entry Points
- [file]: [description]

Database Template

# Database: [Type] + [ORM]

## Connection
- Config: [file path]
- Repo/Connection: [module/class name]

## Schema Overview ([N] tables)

### [Entity Group 1]
- table1 (field1, field2, field3)
  - FK: field_id -> other_table
- table2 (...)

### [Entity Group 2]
...

## Relationships
[ASCII diagram or description]

## Key Indexes
- table.field (unique/index)
...

## Commands
- Migrate: `[command]`
- Rollback: `[command]`
- Create migration: `[command]`

Testing Template

# Testing: [Framework]

## Location
- Tests: [path pattern]
- Fixtures: [path]
- Mocks: [path or approach]

## Commands
- Run all: `[command]`
- Run single: `[command]`
- With coverage: `[command]`

## CI/CD
- Platform: [GitHub Actions/GitLab CI/etc]
- Config: [file path]

Overview Template

# [Project Name]

[One paragraph description]

## Stack
- Language: [lang] [version]
- Framework: [framework]
- Database: [db]

## Quick Start
```bash
[clone]
[install]
[run]

Key Commands

• [cmd1]: [description]
• [cmd2]: [description]

### Step 4: Output

After saving memories, show:

🔍 Analisando projeto...

📁 Detectado: Linguagem: [lang] [version] Framework: [framework] Banco: [database]

💾 Memórias salvas: ✓ context:project:overview ✓ architecture:project:structure ✓ implementation:project:stack ✓ implementation:project:database (se existir) ✓ testing:project:testing

🎉 Onboarding completo! O Claude Code agora conhece seu projeto.

Dica: Use "cx memory search 'como testar'" para verificar.

### Step 5: Handle Flags

**If `--refresh`:**
- Check existing memories with `mcp__cortex__memory(action="list")`
- Update content instead of creating duplicates

**If `--only <type>`:**
- Only analyze and save the specified aspect
- Valid types: stack, architecture, database, testing, overview

**If `--dry-run`:**
- Show what would be saved
- Do NOT call `mcp__cortex__memory(action="save")`
- Format output as preview

## Notes

- Be thorough but concise in memory content
- Focus on information useful for coding tasks
- Include actual file paths and commands
- For database, prioritize understanding relationships
- Adapt to the specific stack found