Menu
First-time maina setup and configuration in any AI coding tool.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Basé sur la classification professionnelle SOC
Use maina cloud for team prompt sync, hosted verification, and feedback-driven learning.
Run maina's two-stage AI code review checking spec compliance then code quality.
Generate rich codebase context using maina's 4-layer retrieval with dynamic token budgets.
Scaffold and validate feature plans using maina's spec-first planning workflow.
Follow maina's test-driven development cycle from generated stubs through red-green-refactor.
Run maina's full verification pipeline on staged changes before committing code.
| name | onboarding |
| description | First-time maina setup and configuration in any AI coding tool. |
| triggers | ["setup","configure","onboard","first time","getting started"] |
When setting up maina for the first time in a repository, configuring it for a new AI coding tool, or onboarding a new team member. This skill covers initial setup, tool discovery, and per-tool configuration.
maina init (or npx @mainahq/cli init). This creates .maina/ config, detects your project stack, and writes a constitution file.maina setup. This walks through model selection, API keys, and tool-specific MCP configuration.maina doctor. This checks that the CLI, MCP server, AI provider, and verification tools are all working.maina wiki init. This scans your codebase, extracts entities via tree-sitter, and generates initial wiki articles.| MCP Tool | Description |
|---|---|
getContext | 4-layer context retrieval with dynamic token budget |
getConventions | Project conventions and constitution |
verify | Full verification pipeline on staged changes |
checkSlop | Detect AI-generated filler text and placeholders |
reviewCode | Two-stage AI code review (spec + quality) |
explainModule | Explain a module's purpose, exports, and dependencies |
suggestTests | Generate test stubs from feature plans |
analyzeFeature | Validate spec-plan consistency |
wikiQuery | Search and synthesize answers from wiki articles |
wikiStatus | Wiki health dashboard |
brainstorm -> ticket -> plan -> design -> spec -> implement
|
pr <- commit <- review <- verify
|
merge -> learn -> improve
MCP auto-configured via .claude/settings.json. Run maina init and it is ready.
MCP via .cursor/mcp.json or project-level .mcp.json. Rules loaded from .cursorrules. Run maina init.
Rules loaded from .windsurfrules. MCP requires global configuration: run maina setup to write it.
MCP auto-discovered from .continue/mcpServers/maina.json. Run maina init.
Rules loaded from .clinerules. MCP configured via VS Code settings: run maina setup.
MCP via .roo/mcp.json. Rules loaded from .roo/rules/maina.md. Run maina init.
MCP via .vscode/mcp.json. Instructions loaded from .github/copilot-instructions.md. Run maina init.
MCP via .amazonq/mcp.json. Run maina init.
MCP via global ~/.config/zed/settings.json. Run maina setup to write the config.
No MCP support. Uses CONVENTIONS.md and .aider.conf.yml for context. Run maina init.
MCP via .mcp.json. Instructions loaded from GEMINI.md. Run maina init.
Instructions loaded from AGENTS.md. Run maina init.
# First-time setup
npx @mainahq/cli init
# Detected: TypeScript, bun, biome
# Created .maina/ configuration
# Wrote constitution to .maina/constitution.md
# Configured MCP for Claude Code, Cursor, Copilot
maina doctor
# CLI: OK v1.1.0
# MCP: OK 10 tools registered
# AI: OK OpenRouter (standard: claude-sonnet-4)
# Verify: OK Biome, Semgrep, Trivy detected
# Wiki: WARN Not initialized (run maina wiki init)
maina wiki init
# Scanned 142 files, extracted 891 entities
# Generated 48 wiki articles
# Wiki ready at .maina/wiki/
maina <command> or npx @mainahq/cli <command>) and MCP tools inside AI coding tools.maina init for tools that support project-level config files.maina setup instead.maina doctor at any time to verify your setup is healthy.