Data MCP v1.4

Secure Data Access
for AI Agents

Give your agents the ability to read, search, and update Docyrus records through a secure Model Context Protocol gateway with built-in multi-tenant guardrails.

Browse Tools Configuration Guide

AI Agent

Docyrus MCP AUTH & PROXY

Tenant Data

Core Tools

Enterprise-Ready CRUD

Full lifecycle management for your records, governed by role-based access.

Fetch Records

Retrieve single records or list filtered results with full relationship expansion.

fetchRecord()

Update & Create

Agents can perform upsert operations with automatic field validation and audit trails.

upsertRecord()

Deep Search

Full-text search across data sources with support for complex filters and sorting.

searchRecords()

Audit Logs

Track every agent action with detailed history—know who changed what and when.

fetchHistory()

Security First

Multi-Tenant Safety
by Default

Implicit Tenant Context

The agent never sees data outside its authorized tenant boundaries. Security is baked into the proxy layer.

OAuth2 Scoping

Leverage existing Docyrus roles and permissions to strictly govern what tools an agent can execute.

Structured Responses

No messy HTML or unparsed text. All data returns as clean, validated JSON optimized for LLM consumption.

mcp-config.json

{
  "mcpServers": {
    "docyrus-data": {
      "command": "npx",
      "args": ["@docyrus/data-mcp"],
      "env": {
        "DOCYRUS_ACCESS_TOKEN": "...",
        "DOCYRUS_TENANT_ID": "...",
        "ALLOWED_DATA_SOURCES": ["crm_*", "billing_*"]
      }
    }
  }
}

Q&A

Common Questions

Does the agent have full DB access?

No. Access is strictly controlled through the data source layer and your specified OAuth2 scope. You decide exactly which tables and operations are allowed.

Can I limit the agent to read-only?

Yes. By using a read-only access token or scoping the MCP server correctly, you can ensure agents only perform harmless query operations.

What LLMs are supported?

Docyrus Data MCP follows the open Model Context Protocol standard, making it compatible with Claude, ChatGPT, and any other MCP-compatible host.

Connect Data to
Action

Empower your agents with secure, governed access to the information they need to be productive.

View on GitHub

Secure Data Access for AI Agents