Terry-Form MCP
Execute Terraform commands locally in a secure, containerized environment. Features LSP integration for intelligent Terraform development.
Terry-Form MCP
AI-powered Terraform execution through the Model Context Protocol.
Terry-Form MCP is a containerized Model Context Protocol server that gives AI assistants like Claude safe, structured access to Terraform. It exposes 25 MCP tools spanning Terraform execution, LSP intelligence, GitHub integration, and Terraform Cloud connectivity — all running inside Docker with destructive operations blocked by design.
Dashboard

The built-in web dashboard provides real-time server health monitoring, tool category overview, and integration status at a glance. Live status auto-refreshes every 5 seconds.
Configuration UI

A tabbed configuration interface lets you manage server settings, integrations, cloud provider credentials, and rate limits — all without touching config files. Built with the HAT stack (HTMX + Alpine.js + Tailwind CSS).
| GitHub Integration | Cloud Providers | Rate Limits |
|---|---|---|
![]() | ![]() | ![]() |
Tool Catalog

The interactive tool catalog at /tools lists all 25 MCP tools with search, category filtering, and expandable parameter details. Also available as a raw JSON endpoint at /api/tools and as a static tools.json file.
Quick Start
Prerequisites
- Docker installed and running
- Python >= 3.10 (for local development)
1. Build
scripts/build.sh # Linux/macOS
scripts\build.bat # Windows
# or directly:
docker build -t terry-form-mcp .
2. Run as MCP Server
docker run -it --rm \
-v "$(pwd)":/mnt/workspace \
terry-form-mcp
3. Verify the Image
scripts/verify.sh # Runs 8 checks: Docker, image size, Terraform, terraform-ls, Python, files, tools, startup
MCP Client Configuration
Add Terry-Form to any MCP-compatible client:
{
"mcpServers": {
"terry": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-v", "/path/to/your/workspace:/mnt/workspace",
"terry-form-mcp"
]
}
}
}
Platform-specific examples
Claude Desktop (Windows)
{
"mcpServers": {
"terry": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-v", "C:\\Users\\YourUsername\\terraform-projects:/mnt/workspace",
"terry-form-mcp"
]
}
}
}
Claude Desktop (macOS)
{
"mcpServers": {
"terry": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-v", "/Users/YourUsername/terraform-projects:/mnt/workspace",
"terry-form-mcp"
]
}
}
}
VSCode (uses workspace variable)
{
"mcp.servers": {
"terry": {
"command": "docker",
"args": [
"run", "-i", "--rm",
"-v", "${workspaceFolder}:/mnt/workspace",
"terry-form-mcp"
]
}
}
}
Tools (25)
| Category | Tools | Count |
|---|---|---|
| Core Terraform | terry, terry_version, terry_environment_check, terry_workspace_list | 4 |
| LSP Intelligence | terraform_validate_lsp, terraform_hover, terraform_complete, terraform_format_lsp, terraform_lsp_status | 5 |
| Diagnostics | terry_lsp_debug, terry_workspace_info, terry_lsp_init, terry_file_check, terry_workspace_setup, terry_analyze | 6 |
| Security | terry_security_scan, terry_recommendations | 2 |
| GitHub | github_clone_repo, github_list_terraform_files, github_get_terraform_config, github_prepare_workspace | 4 |
| Terraform Cloud | tf_cloud_list_workspaces, tf_cloud_get_workspace, tf_cloud_list_runs, tf_cloud_get_state_outputs | 4 |
Core Terraform
# Initialize and validate a project
terry(path="infrastructure/aws", actions=["init", "validate"])
# Plan with variables
terry(path="environments/prod", actions=["plan"], vars={"instance_count": "3", "region": "us-east-1"})
Only init, validate, fmt, and plan are permitted. apply and destroy are blocked.
LSP Intelligence
# Code completions
terraform_complete(file_path="main.tf", line=10, character=0)
# Hover documentation
terraform_hover(file_path="main.tf", line=15, character=12)
# Detailed validation with error locations
terraform_validate_lsp(file_path="main.tf")
# Format a file
terraform_format_lsp(file_path="main.tf")
Powered by terraform-ls v0.38.5 — provides context-aware completions, inline documentation, and diagnostics with precise source locations.
GitHub Integration
# Clone a repo and prepare it for Terraform operations
github_clone_repo(owner="myorg", repo="infrastructure")
github_prepare_workspace(owner="myorg", repo="infrastructure", config_path="environments/prod")
Security Scanning
# Scan for hardcoded credentials, missing encryption, overly permissive policies
terry_security_scan(path="my-project")
# Get actionable improvement recommendations
terry_recommendations(path="my-project")
Architecture
┌─────────────┐ MCP Protocol ┌──────────────────────────────────────┐
│ AI Assistant │ ◄──────────────────► │ Terry-Form MCP Server │
│ (Claude) │ │ │
└─────────────┘ │ ┌─────────────┐ ┌──────────────┐ │
│ │ Terraform │ │ terraform-ls │ │
│ │ CLI 1.12 │ │ LSP 0.38.5 │ │
│ └──────┬───────┘ └──────┬───────┘ │
│ │ │ │
│ ▼ ▼ │
│ ┌──────────────────────────────┐ │
│ │ /mnt/workspace (isolated) │ │
│ └──────────────────────────────┘ │
└──────────────────────────────────────┘
Docker Container
Key Components
| File | Purpose |
|---|---|
src/server_enhanced_with_lsp.py | Main FastMCP server — registers all 25 tools |
src/terry-form-mcp.py | Core Terraform subprocess execution |
src/terraform_lsp_client.py | Async LSP client wrapping terraform-ls |
src/mcp_request_validator.py | Input sanitization, path traversal prevention, rate limiting |
src/github_repo_handler.py | Clone repos and extract Terraform files |
src/github_app_auth.py | GitHub App JWT/OAuth authentication |
src/frontend/ | HAT stack web UI (dashboard + configuration) |
Frontend Stack
The built-in web UI uses the HAT stack:
- HTMX 2.0 — partial page updates without full reloads
- Alpine.js 3.14 — lightweight client-side reactivity for tabs and toasts
- Tailwind CSS — dark-mode-first utility styling
Accessible at the server root when running with streamable-http or sse transport.
Security Model
Terry-Form implements defense-in-depth with four layers:
| Layer | Protection |
|---|---|
| Container Isolation | All execution in ephemeral Docker containers. No host access. |
| Operation Allowlist | Only init, validate, fmt, plan. No apply/destroy. |
| Workspace Isolation | All file operations restricted to /mnt/workspace. Path traversal blocked. |
| Input Validation | JSON schema enforcement, variable sanitization, rate limiting per category. |
Forced environment variables: TF_IN_AUTOMATION=true, TF_INPUT=false, CHECKPOINT_DISABLE=true.
Running with the Web UI
To use the dashboard and configuration UI, run with HTTP transport:
# Local
MCP_TRANSPORT=streamable-http HOST=0.0.0.0 PORT=8000 python3 src/server_enhanced_with_lsp.py
# Docker
docker run -it --rm \
-p 8000:8000 \
-v "$(pwd)":/mnt/workspace \
-e MCP_TRANSPORT=streamable-http \
terry-form-mcp
Then open http://localhost:8000 in your browser.
Configuration Tabs
| Tab | What it configures |
|---|---|
| Server | Transport mode, host, port, API key |
| GitHub | App ID, installation ID, private key path, webhook secret |
| Terraform Cloud | API token |
| Cloud Providers | AWS, GCP, and Azure credentials |
| Rate Limits | Per-category request limits (applied immediately) |
| Terraform Options | Log level, operation timeout |
Container Details
Built on hashicorp/terraform:1.12 (Alpine-based, ~150MB). Includes:
- Terraform CLI 1.12
terraform-lsv0.38.5 for LSP support- Python 3.12 with FastMCP 3.0+
- Runs as non-root user
terraform(UID 1001)
Development
# Install dependencies
pip install -r requirements.txt
# Run locally
python3 src/server_enhanced_with_lsp.py
# Code quality
black . # Format (88-char line limit)
flake8 . # Lint
mypy src/*.py # Type check
Limitations
- No state modification —
applyanddestroyare intentionally blocked - String variables only — complex variable types not supported via CLI passthrough
- LSP cold start — first LSP operation takes 1-2 seconds for initialization
- Local execution — designed for development workflows, not production CI/CD
License
相關伺服器
Scout Monitoring MCP
贊助Put performance and error data directly in the hands of your AI assistant.
Alpha Vantage MCP Server
贊助Access financial market data: realtime & historical stock, ETF, options, forex, crypto, commodities, fundamentals, technical indicators, & more
Email MCP Server by Sidemail
Let AI agents write & manage your SaaS emails
MCP Tool Kit
A modular server for building AI agents with a unified interface for tools like file operations, web search, browser automation, and financial data.
Chrome DevTools MCP
chrome-devtools-mcp lets your coding agent (such as Gemini, Claude, Cursor or Copilot) control and inspect a live Chrome browser
shadcn/ui
Provides structured data for shadcn/ui components, including descriptions, installation instructions, usage examples, and props.
secretctl
AI-safe secrets manager - inject credentials as env vars, AI never sees plaintext
GameCode MCP2
A Model Context Protocol (MCP) server for tool integration, configured using a tools.yaml file.
Jenkins MCP Server
An MCP server for automating tasks and managing jobs on a Jenkins server.
@mcp-fe/react-tools
Don't let AI guess from screenshots. Give LLMs direct access to your React state, Context, and Data Grids. Features bidirectional communication via SharedWorkers & WebSockets. Docker gateway included.
Unity Editor MCP
Enables AI assistants to interact directly with the Unity Editor for AI-assisted game development and automation.
Postman MCP Server
Run Postman collections using Newman, with support for environment and global variables.


