tf_code/docs/sync-implementation.md

tfcode Sync Implementation Summary

What's Implemented

1. Postinstall Script (packages/tfcode/scripts/postinstall.cjs)

• Checks for Python 3.10+ installation
• Auto-installs ToothFairyAI Python SDK on npm install tfcode
• Handles externally-managed environments (Homebrew, etc.)
• Shows clear error messages if Python is missing

2. Python Sync Module (packages/tf-sync/src/tf_sync/)

• config.py - Configuration management, credential validation, multi-region support
• tools.py - Sync API functions from TF workspace
• mcp.py - Reserved for future MCP support (not in SDK yet)

3. CLI Commands (packages/tfcode/src/cli/cmd/tools.ts)

• tfcode validate - Validate TF credentials
• tfcode sync - Sync tools from workspace
• tfcode tools list [--type] - List synced tools
• tfcode tools credentials <name> --set/--show - Manage API keys
• tfcode tools debug <name> - Debug tool configuration

4. Region Support

Region	Base URL	Use Case
dev	api.toothfairylab.link	Development
au	api.toothfairyai.com	Australia (Default)
eu	api.eu.toothfairyai.com	Europe
us	api.us.toothfairyai.com	United States

Test Results

$ python3 scripts/test-sync.py

============================================================
tfcode Sync Test
============================================================

Test 1: Load Configuration
----------------------------------------
✓ Workspace ID: 6586b7e6-683e-4ee6-a6cf-24c19729b5ff
✓ Region: dev
✓ API URL: https://api.toothfairylab.link

Test 2: Validate Credentials
----------------------------------------
✓ Credentials valid
  Workspace: Connected
  ID: 6586b7e6-683e-4ee6-a6cf-24c19729b5ff

Test 3: Sync Tools
----------------------------------------
✓ Synced 100 tools

By type:
  api_function: 100

Sample tools:
  🌐 get_kanban_projects (api_function)
  🌐 azure_costs_test_official (api_function)
  ... and 95 more

SDK Capabilities (Current)

The ToothFairyAI Python SDK currently exposes:

• agent_functions - API Functions with request_type (get/post/put/delete/etc.)
• connections - Provider connections (openai, anthropic, groq, etc.)
• agents - TF workspace agents

Not yet exposed:

• MCP servers (isMCPServer)
• Agent skills (isAgentSkill)
• Database scripts (isDatabaseScript)

These will be added to the SDK in the future. For now, MCP servers must be configured manually in tfcode.json.

Environment Setup

export TF_WORKSPACE_ID="your-workspace-id"
export TF_API_KEY="your-api-key"
export TF_REGION="dev"  # or au, eu, us

Usage

# Install
npm install -g tfcode

# Or with bun
cd packages/tf-sync && python3 -m pip install -e . --break-system-packages
cd packages/tfcode && bun install

# Validate
tfcode validate

# Sync
tfcode sync

# List tools
tfcode tools list

Files Created/Modified

• packages/tfcode/scripts/postinstall.cjs - Postinstall script
• packages/tfcode/src/cli/cmd/tools.ts - CLI commands
• packages/tf-sync/src/tf_sync/config.py - Config + regions
• packages/tf-sync/src/tf_sync/tools.py - Tool sync
• packages/tf-sync/src/tf_sync/mcp.py - MCP placeholder
• scripts/test-sync.py - Test script
• scripts/setup-tf-dev.sh - Dev environment setup