AgentSkillsCN

onboard

为全新的 Claude 会话加载 CBass 项目上下文。当您开始新工作、切换关注领域,或对代码库尚不熟悉时,可使用此技能。它将为您提供架构概览、服务清单,以及当前项目的整体状态。

SKILL.md
--- frontmatter
name: onboard
description: Load CBass project context for a fresh Claude session. Use when starting work, switching focus areas, or when unfamiliar with the codebase. Provides architecture overview, service inventory, and current project state.
user-invocable: true
allowed-tools: Read, Glob, Grep

CBass Project Onboarding

Quickly bring a fresh Claude agent up to speed on the CBass self-hosted AI stack.

Usage

code
/onboard           # Full project context
/onboard n8n       # Focus on n8n workflows and automation
/onboard flowise   # Focus on Flowise chatflows and agents
/onboard deploy    # Focus on deployment and operations
/onboard data      # Focus on data layer (Supabase, Qdrant, Neo4j)

What This Skill Provides

1. Project Overview

  • CBass is a self-hosted AI Docker Compose orchestration platform
  • Fork of n8n's self-hosted-ai-starter-kit
  • Educational platform for teaching AI tools (primary learner: biology student)
  • Live deployment at cbass.space

2. Architecture Summary

code
Internet → Caddy (:443) → Internal Services
                              ↓
┌─────────────────────────────┼─────────────────────────────┐
│ Core               │ Data Layer        │ Utilities       │
├────────────────────┼───────────────────┼─────────────────┤
│ n8n (:5678)        │ Supabase (:8000)  │ Langfuse (:3000)│
│ Flowise (:3001)    │ Qdrant (:6333)    │ SearXNG (:8081) │
│ Open WebUI (:8080) │ Neo4j (:7474)     │ Kali (:6901)    │
│ Ollama (:11434)    │ PostgreSQL (:5432)│ Caddy (:80/443) │
└────────────────────┴───────────────────┴─────────────────┘

3. Key Files

FilePurpose
CLAUDE.mdClaude Code instructions and known issues
README.mdProject overview and quick start
docker-compose.ymlService definitions
start_services.pyMain entry point
CaddyfileReverse proxy routing
.envEnvironment variables (secrets - gitignored)

4. Documentation Structure

code
docs/
├── README.md           # Documentation index
├── getting-started/    # First-time setup guides
├── deployment/         # Local and VPS deployment
├── services/           # Per-service documentation (11 services)
├── architecture/       # System design and diagrams
├── operations/         # Day-to-day tasks
└── archive/            # Deprecated docs

Focus Areas

n8n Focus (/onboard n8n)

Read these files:

  • docs/services/n8n.md - n8n service documentation
  • n8n/backup/workflows/ - Pre-built RAG workflows
  • n8n_pipe.py - Open WebUI integration

Key concepts:

  • Webhook triggers for external integration
  • Credential configuration (use container names: ollama, db, qdrant)
  • Known credential IDs in CLAUDE.md

Flowise Focus (/onboard flowise)

Read these files:

  • docs/services/flowise.md - Flowise service documentation
  • flowise/ - Pre-built tools and chatflows
  • wrap_flowise.ps1 - Import format converter

Key concepts:

  • ExportData format vs raw workflow format
  • Settings > Load Data for reliable imports
  • Named volume for data persistence

Deployment Focus (/onboard deploy)

Read these files:

  • docs/deployment/vps-setup.md - Production deployment
  • docs/deployment/local-dev.md - Local development
  • docs/deployment/dns-ssl.md - DNS and SSL setup

Key concepts:

  • GPU profiles: gpu-nvidia, gpu-amd, cpu, none
  • Environment modes: private (dev), public (prod)
  • Caddy handles automatic HTTPS

Data Layer Focus (/onboard data)

Read these files:

  • docs/services/supabase.md - PostgreSQL + pgvector
  • docs/services/qdrant.md - Vector database
  • docs/services/neo4j.md - Graph database

Key concepts:

  • Supabase: Use container name db for PostgreSQL
  • Qdrant: Vector dimensions must match embedding model
  • Neo4j: Auth format is neo4j/password

Current Project State

Check CLAUDE.md for:

  • Current todo items
  • Known issues and workarounds
  • Credential IDs for CBass instance

Quick Reference

Start Services

bash
python start_services.py --profile gpu-nvidia --environment private

Check Status

bash
docker compose -p localai ps

View Logs

bash
docker compose -p localai logs -f <service-name>

Service URLs (Local)

Additional Context Files

For deeper understanding: