Terraform MCP Server: Let AI Check State, Plan Diffs, and Query Resources

Infrastructure State Is Scattered

You write Terraform in your editor, but the state lives in Terraform Cloud. Plan outputs, resource inventories, workspace configs, run history. When your AI assistant needs to understand what's deployed, you switch tabs, find the workspace, dig through run history, and paste the relevant bits.

An MCP server gives your AI direct access to Terraform Cloud operations. Check state, review plans, query resources, and monitor runs through natural language.

How It Works with DataFaucet

DataFaucet captures Terraform Cloud's web API calls as you browse. Those workspace, run, and state operations become MCP tools your AI can invoke.

Step 1: Enter Your Terraform Cloud URL

Go to DataFaucet and paste:

https://app.terraform.io

Step 2: Browse Your Workspaces

Navigate through features you want AI access to:

• Workspaces — open 2-3 workspaces, check their current state
• Runs — view recent runs, check plan/apply status
• State — browse the state file viewer for a workspace
• Variables — check workspace variables and variable sets
• Resources — browse the resource list in a workspace state

Step 3: Deploy and Connect

Review captured endpoints, deploy your server, connect to Claude, Cursor, or any MCP client.

What Your AI Can Do

Once connected:

• "What resources are in the production workspace?"
• "Did the last Terraform run succeed or fail?"
• "Show me the plan diff for the pending run"
• "What variables are set in the staging workspace?"
• "Which workspaces have pending changes?"
• "How many EC2 instances are in the current state?"

Use Cases by Role

Platform engineers: "Which workspaces haven't had a successful run in over a week?" Identify stale infrastructure.

DevOps: "Show me the plan output for the latest run in network-prod. Any destructive changes?" Quick review before approval.

SREs: "What resources exist in the disaster-recovery workspace? Is the backup RDS instance still there?" State verification during incidents.

Tech leads: "Compare the resource count between staging and production workspaces." Drift detection from chat.

Terraform + AI Workflow

Typical infrastructure review:

1. PR changes Terraform config
2. Ask AI: "What does the plan output look like for the latest run?"
3. AI shows additions, changes, and destructions
4. "Are there any destructive changes I should be worried about?"
5. AI highlights resource deletions
6. Approve or request changes with full context

No more reading dense plan diffs in the Terraform Cloud UI.

Tips for Better Capture

1. Workspaces: Browse at least 2 workspaces to capture list and detail endpoints
2. Runs: Open a completed run to capture plan/apply output viewing
3. State: Browse the state viewer to capture resource listing
4. Variables: Open the variables tab to capture variable access
5. Settings: Check workspace settings for config endpoints

Security

DataFaucet captures your authenticated Terraform Cloud session. The MCP server can only access workspaces and organizations your account has permission to view. Team access and workspace permissions are respected.

Your API key protects the MCP endpoint. Sensitive variables (marked as sensitive in Terraform) remain redacted.

Connect in 60 Seconds

1. Go to DataFaucet
2. Paste https://app.terraform.io
3. Browse workspaces, runs, and state (2-3 minutes)
4. Deploy and connect to Claude, Cursor, or Codex

Your infrastructure state becomes AI-queryable. Workspace status, plan diffs, and resource inventory through natural language.

Related: AWS MCP Server for cloud console, Docker MCP Server for containers, Vercel MCP Server for deployments, Datadog MCP Server for monitoring.