Environment-aware setup system that automatically detects your context (VPS, Codespaces, local dev) and installs only the tools you need.

Claude Code + Cline + 10+ Coding Tools

Referral Code: DVARAUG0U8

git clone https://github.com/ebowwa/seed.git
cd seed
./setup.sh

# .env file
AI_ASSISTANT=zai
ZAI_API_KEY=your-key
GITHUB_TOKEN=ghp_xxx
DOPPLER_TOKEN=dp.st.xxx
TAILSCALE_AUTH_KEY=tskey-xxx

# GitHub CLI
gh auth login

# Tailscale
sudo tailscale up  # Linux
tailscale up       # macOS

# Doppler
doppler login

# Run Claude Code with Z.ai config from Doppler
doppler run --project seed --config prd -- claude

The chat.sh wrapper enables persistent conversations with Claude across sessions.

# Basic usage
./chat.sh "your prompt here"

# Custom project/config
./chat.sh "prompt" --project myproj --config dev

# Via environment variables
DOPPLER_PROJECT=other DOPPLER_CONFIG=staging ./chat.sh "prompt"

How it works:

• Maintains conversation history in /tmp/c.txt
• Each prompt includes full conversation context
• Loads secrets from Doppler (configurable)
• Resets when /tmp/c.txt is deleted

Communicate with a remote Claude instance on a VPS for distributed AI workflows.

# Example: Chat with remote Claude on Hetzner server
ssh ebowwa-deptwar "cd ~/seed && ./chat.sh 'your prompt'"

Use cases:

• Different backends (Z.ai GLM vs Anthropic)
• Independent context/persistence
• Multi-Claude collaboration
• Test prompts across models without local config changes

• setup.sh - Main installation script
• situations.yaml - Environment/tool configs
• chat.sh - Persistent conversation wrapper
• .env.example - API key template
• .claude/settings.template.json - Claude settings template
• .claude/settings.local.json - Local Claude settings
• .claude/commands/ - Claude Code skills (see distributed-claude.md)

• Claude Code Skills
• GLM Models Comparison
• GitHub MCP Integration
• Z.ai Vision MCP
• Z.ai Web Search MCP

• Never commit .env files
• Rotate credentials regularly
• Use environment-specific tokens

MIT