Enterprise MCP Gateway
Securely bridge AI agents to internal REST APIs with RBAC, PII masking, and audit logging via MCP.
Free
Works with the AI Agents you already use
Enterprise MCP Gateway
Example session with this skill installed
Help me connect my Claude Desktop agent to our internal inventory API using the enterprise gateway. The API spec is at /opt/specs/inventory.yaml and my token is 'dev-token-99'.
- Read your context and instructions
- Compiled the enterprise mcp gateway
I have generated the configuration files required to bridge your internal inventory API to Claude Desktop via the Enterprise MCP Gateway. This setup ensures all calls are logged and PII is redacted before reaching the agent.
Gateway Configuration and Connection
First, create your config.yaml to define the connector and security rules:
# config.yaml
connectors:
- name: inventory-api
spec_file: "/opt/specs/inventory.yaml"
base_url: "http://internal-inventory.local:8081"
governance:
mask_secrets: true
tokens:
dev-token-99:
role: developer
roles:
developer:
allowed_tools:
- "inventory-api:list_*"
- "inventory-api:get_*"
sanitizer:
sensitive_keys:
- "api_key"
- "session_id"
- "credit_card"
audit:
log_path: "logs/mcp-audit.json"
format: "json"
Next, update your claude_desktop_config.json to include the gateway:
{
"mcpServers": {
"enterprise-gateway": {
"command": "/usr/local/bin/mcp-gateway",
"args": [
"--config", "/absolute/path/to/config.yaml",
"--token", "dev-token-99"
]
}
}
}
Next steps
- Ensure the
mcp-gatewaybinary has execution permissions (chmod +x). - Verify the
base_urlis reachable from your local machine. - Restart Claude Desktop to initialize the MCP tools.
Connects securely to your tools. The creator never sees your data.
About this skill
Exposing internal enterprise APIs to AI agents often creates security and compliance gaps. This skill provides a secure Model Context Protocol (MCP) gateway that bridges AI coding agents to internal REST and OpenAPI services with enterprise-grade controls. It ensures that internal tools are only accessible via governed channels, preventing unauthorized access and data leakage.
What it does
- OpenAPI bridging connects agents like Claude, Cursor, and Gemini to internal REST services by converting OpenAPI 3.0 specs into MCP tools.
- RBAC enforcement applies granular tool-level permissions per agent or team using token-based role mapping.
- Data redaction masks sensitive info including credit cards, SSNs, and secrets using zero-allocation in-flight PII filtering.
- Audit logging generates immutable JSON logs of every tool execution for SIEM compliance and security forensics.
- Flexible transport supports local execution via
stdiofor IDE agents or centralized deployment viaSSEfor shared team servers.
How it works
- Define connectors in a
config.yamlfile pointing to absolute paths of OpenAPI 3.0 specification files. - Configure governance by mapping tokens to roles and defining allowed tool globs and PII masking patterns.
- Launch the binary using the
--configflag and selecting the transport mode (stdioorsse). - Connect the agent by adding the gateway command and token to the agent's MCP configuration file.
Frameworks & tools
The gateway is built as a Go binary and supports OpenAPI 3.0 specifications. It integrates with Claude Code, Claude Desktop, Cursor, Antigravity, and Gemini CLI.
Why this beats prompting it yourself
Prompting an agent to call an API manually lacks security controls and auditability. This skill moves governance to the infrastructure layer, ensuring PII is redacted and access is logged before the LLM ever sees the data.
Use cases
- Providing developers with AI-driven access to internal CRM or ERP APIs while masking customer PII.
- Restricting a support agent to read-only tool access while allowing an admin agent full write capabilities.
- Maintaining a centralized audit trail of all actions performed by AI agents against internal production environments.
Known limitations
The gateway currently only supports OpenAPI 3.0 REST specifications. GraphQL and gRPC backends are not supported. Config files must use absolute paths to resolve OpenAPI specs correctly.
Supports OpenAPI 3.0 REST only. No support for GraphQL or gRPC. Requires absolute file paths in configuration. Audit logs must be routed to stderr or file in stdio mode to avoid stream pollution.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Requires
Permissions
Allowed Hosts