Ecosystem & Marketplaces›Open-Source MCP Servers (10)›Terraform MCP Server›Tools

Tools reference

The Terraform MCP Server exposes 3 core tools covering sandboxed Terraform CLI execution, vector similarity search across provider documentation, and multi-format documentation ingestion.

Tool	Category	Description
terraform_execute	Execution	Execute whitelisted Terraform commands (init, validate, plan, apply, destroy) inside validated working directories
terraform_doc_search	Semantic Search	Query Neo4j HNSW vector indexes for Terraform provider resource schemas, data sources, and best practices
ingest_terraform_docs	Ingestion	Parse, chunk, embed, and store Terraform provider documentation into Neo4j

1. terraform_execute

Runs whitelisted Terraform CLI commands with directory traversal protection, variable injection scanning, timeout enforcement, and ANSI output sanitization.

Supported subcommands

Command	Behavior	Auto-Approve Flag
init	Initializes the working directory and downloads provider plugins	No
validate	Validates HCL syntax and internal resource consistency	No
plan	Generates an execution plan previewing resource additions, changes, and deletions	No
apply	Applies the planned infrastructure changes	Yes (-auto-approve)
destroy	Destroys managed infrastructure in the working directory	Yes (-auto-approve)

Parameters

Parameter	Type	Required	Default	Description
command	string	Yes	—	Terraform subcommand (init, plan, validate, apply, destroy)
working_directory	string	Yes	—	Absolute path to the directory containing .tf files
variables	dict[str, str]	No	None	Key-value pairs passed as -var arguments (max 100 keys)
aws_region	string	No	None	Optional AWS region override (AWS_DEFAULT_REGION)
strip_ansi	boolean	No	True	Strip ANSI escape sequences from stdout/stderr
timeout	integer	No	300	Command execution timeout in seconds (1–1800)

Response structure

{
  "success": true,
  "data": {
    "command": "terraform plan",
    "status": "success",
    "result": {
      "return_code": 0,
      "stdout": "Terraform will perform the following actions...",
      "stderr": "",
      "execution_time": 2.45
    },
    "metadata": {
      "terraform_version": "1.9.5",
      "aws_region": "us-west-2",
      "variables_count": 2,
      "security_checks_passed": true
    }
  }
}

2. terraform_doc_search

Performs semantic similarity search over ingested Terraform documentation using Neo4j HNSW vector indexes and cosine similarity scoring.

Parameters

Parameter	Type	Required	Default	Description
query	string	Yes	—	Natural language or resource schema search query (1–1000 characters)
top_k	integer	No	5	Maximum number of chunks to return (1–50)
similarity_threshold	float	No	0.7	Minimum cosine similarity score (0.0–1.0)
node_types	list[str]	No	None	Filter by document node type: resource, data_source, best_practice

Indexed node types

Node Type	Neo4j HNSW Index	Contents
resource	docchunk_resource_embedding_hnsw	Provider resource arguments, nested blocks, and attribute exports
data_source	docchunk_datasource_embedding_hnsw	Provider data source lookup schemas and filter blocks
best_practice	docchunk_bestpractice_embedding_hnsw	Architecture patterns, state management guidelines, and security baselines

Response structure

{
  "success": true,
  "data": {
    "query": "AWS S3 bucket server-side encryption configuration",
    "results_count": 3,
    "results": [
      {
        "content": "resource \"aws_s3_bucket_server_side_encryption_configuration\" \"example\" {...}",
        "similarity_score": 0.91,
        "node_type": "resource",
        "id": "docchunk_s3_sse_001"
      }
    ],
    "search_parameters": {
      "top_k": 5,
      "similarity_threshold": 0.7
    }
  }
}

3. ingest_terraform_docs

Ingests Markdown, HTML, or PDF Terraform provider documentation into the Neo4j knowledge graph with semantic chunking and vector embedding generation.

Parameters

Parameter	Type	Required	Default	Description
filter_types	list[str]	Yes	—	Document types to ingest: resource, data_source, terraform, best_practice, readme
filter_services	list[str]	No	None	Optional cloud service filter (for example, ["EC2", "S3", "VPC"])
scan_dirs	list[str]	No	["docs/"]	Local filesystem directories to scan for documentation files

Response structure

{
  "success": true,
  "data": {
    "ingestion_summary": {
      "total_documents": 150,
      "successful": 148,
      "failed": 2,
      "skipped": 0
    },
    "types_processed": ["resource", "data_source"],
    "chunks_created": 1250,
    "embeddings_generated": 1250
  }
}