ENTIA MCP Server

Structured business intelligence for AI agents.

ENTIA provides verified entity data across 34 countries — accessible via Model Context Protocol (MCP) or REST API.

Quick Start (< 2 minutes)

Option 1: Remote MCP Server (recommended)

No installation needed. Connect your MCP client directly:

Claude Desktop — add to claude_desktop_config.json:

{
  "mcpServers": {
    "entia": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.entia.systems/mcp"]
    }
  }
}

Cursor IDE — add to .cursor/mcp.json:

{
  "mcpServers": {
    "entia": {
      "command": "npx",
      "args": ["mcp-remote", "https://mcp.entia.systems/mcp"]
    }
  }
}

Then try:

Look up Telefonica in Spain

Option 2: REST API

# Search entities
curl "https://entia.systems/v1/search?q=telefonica&country=ES&limit=5" \
  -H "X-ENTIA-Key: YOUR_API_KEY"

# Full entity profile (BORME + GLEIF + VIES + Wikidata)
curl "https://entia.systems/v1/profile/Telefonica?country=ES"

# EU VAT verification
curl "https://entia.systems/v1/verify/vat/ESA28015865"

# Platform stats
curl "https://entia.systems/v1/stats"

Option 3: Python SDK (coming soon)

A Python client (entia-mcp on PyPI) and LangChain integration are on the roadmap. Not yet published. Until then, use Option 1 (MCP) or Option 2 (REST).

6 MCP Tools

Pricing

Get your API key: entia.systems/mcp-setup

Data Sources

All data comes from official public registries:

• BORME -- Spanish Mercantile Registry (BOE)

• VIES -- EU VAT validation (European Commission)

• GLEIF -- Legal Entity Identifiers (Global LEI Foundation)

• Wikidata -- Knowledge Graph (Wikimedia Foundation)

• REPS -- Spanish Healthcare Professionals Registry

• INE -- Spanish National Statistics Institute

• SEPE -- Spanish Employment Service

• AEAT -- Spanish Tax Authority

• Companies House -- UK company registry

• Sirene/INSEE -- French company registry

Links

• API Documentation

• Get API Key

• Setup Guide

• Client Dashboard

• Official MCP Registry

About

Built by PrecisionAI Marketing OU (Estonia, EU).

• VAT: EE102780516

• DUNS: 565868914

• e-Residency certified

• eIDAS compliant

License

Proprietary. See Terms of Service.

AWS deployment (emergency migration runbook)

This repository now includes a minimal AWS bootstrap for running the standalone MCP server on ECR + ECS Fargate + ALB + Route 53 + ACM + Secrets Manager + CloudWatch Logs under infra/aws/.

What is deployed

• ECR repository for container images.

• ECS cluster, task definition, and Fargate service.

• ALB with:

  • port 80 redirect to 443

  • port 443 HTTPS listener (ACM certificate)

  • target group health check on /health

• Security groups (ALB public ingress, ECS ingress from ALB only).

• Route53 alias record (optional) for mcp.entia.systems.

• CloudWatch Logs group for ECS logs.

• IAM roles with least-privilege baseline, plus optional Secrets Manager access.

Runtime assumptions preserved

• MCP_TRANSPORT=http

• MCP_PORT=3000

• Health endpoint path: /health

Required AWS environment variables (GitHub Actions repository variables)

• AWS_REGION

• ECR_REPOSITORY (example: 123456789012.dkr.ecr.us-east-1.amazonaws.com/entia-mcp-prod-mcp)

• ECS_CLUSTER

• ECS_SERVICE

• DOCKERFILE_PATH (optional, defaults to Dockerfile)

Required GitHub Actions secrets

• AWS_GITHUB_ACTIONS_ROLE_ARN (OIDC role ARN with Terraform/ECR/ECS permissions)

• TF_VARS_JSON_B64 (base64-encoded JSON tfvars for required Terraform inputs except container_image)

Required runtime secrets (AWS Secrets Manager)

Configure as secret_arns map in Terraform (ENV_VAR_NAME => secret ARN), for example:

• ENTIA_API_KEY

• Any upstream API credentials used by the standalone MCP server

• SMTP or provider tokens if used by the runtime

Keep all credentials in Secrets Manager. Do not hardcode secrets in workflow files or Terraform variables.

Deploy steps

1. Copy infra/aws/terraform.tfvars.example to infra/aws/terraform.tfvars.

2. Fill network/DNS/certificate/image values.

3. Run:

   • terraform -chdir=infra/aws init

   • terraform -chdir=infra/aws plan

   • terraform -chdir=infra/aws apply

4. For CI deploys, provide TF_VARS_JSON_B64 secret (base64 of tfvars JSON) and trigger workflow AWS MCP Deploy.

5. Push to main or trigger GitHub Actions workflow AWS MCP Deploy.

6. Validate smoke checks:

   • GET https://mcp.entia.systems/health returns 200

   • MCP endpoint is reachable via ALB

   • ECS service has healthy targets

Rollout steps (production-safe)

1. Deploy infra and ECS service without changing DNS (create_dns_record=false) and test on ALB DNS name.

2. Run smoke checks on ALB hostname.

3. Enable DNS alias (create_dns_record=true) once healthy.

4. Monitor CloudWatch Logs + ALB target health for at least one full business cycle.

Rollback steps

1. Revert DNS alias to previous serving endpoint (mcp.entia.systems).

2. Re-deploy previous stable image tag or previous task definition revision in ECS:

   • aws ecs update-service --cluster <cluster> --service <service> --task-definition <family:revision> --force-new-deployment

3. If needed, roll back Terraform to previous commit and apply.

4. Confirm /health and MCP endpoint recovery.

DNS cutover steps for mcp.entia.systems

1. Ensure ACM cert is issued for mcp.entia.systems in the deployment region.

2. Apply Terraform with create_dns_record=true, hosted_zone_id, and domain_name="mcp.entia.systems".

3. Verify Route53 alias points to ALB and target group is healthy.

4. Execute post-cutover checks:

   • TLS certificate chain valid

   • /health = 200

   • MCP request/response path works end-to-end

For additional safety/rollback details, see AGENTS.md and docs/aws-migration-assumptions.md.