> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.merge.dev/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.merge.dev/_mcp/server.

> Agent Handler is the production layer between your AI agent and the systems it acts on: hundreds of Connectors, OAuth and credential management per end user, a Security Gateway that scans every tool input and output, and observability over every call.

<h1>
  Merge Agent Handler
</h1>

<p>
  Merge Agent Handler is the tool-calling platform for AI agents. It gives your agent secure, reliable access to the tools your users need - across hundreds of MCP-ready third-party services.
</p>

<p>
  Authentication per end user or shared across a Group, tool scoping per agent surface, data-loss prevention on every call, and a complete Audit Trail. Enterprise-ready out of the box.
</p>

<h2>
  Features
</h2>

#### Hundreds of connectors

Pre-built MCP-ready connectors for the SaaS your customers use - Salesforce, Slack, Jira, GitHub, HubSpot, NetSuite, Workday, and the rest. Tool schemas are normalized; descriptions are model-friendly.

#### Auth per end user

Per-user authentication or shared credentials across a Group, depending on whether each end user has their own third-party account or a tenant shares one. OAuth, API keys, and custom auth supported. Bring your own OAuth app for branding on the consent screen.

#### Tool Packs

A bundle of tools that determines what an agent can act on at a given point in time. Used to define tool access for an individual or an agent. Each pack is a bounded surface area you can govern, audit, and update independently.

#### Security Gateway

Every tool call is scanned for PII, PHI, payment data, and any custom regex you define before it leaves Agent Handler for the third party. Configure allow, redact, or block per entity, with overrides per Tool Pack.

#### Observability

Every tool call is captured: arguments, result, latency, the user it ran for, redactions applied, errors. Filter by user or tool. Audit log covers admin actions. Webhooks fan events out to your own pipeline.

#### Enterprise governance

OAuth-authenticated MCP server, SSO and SCIM for your team, custom roles, audit-trail export. The controls your security review expects, set up correctly from day one.

<h2>
  Start here
</h2>

#### [Quickstart](/merge-agent-handler/quickstart)

Connect Agent Handler to your IDE and make your first tool call in about five minutes.

#### [How it works](/merge-agent-handler/how-it-works)

The architecture, top to bottom. What MCP is, what Agent Handler adds, where the Security Gateway sits, what the audit log captures.

#### [Use cases](/merge-agent-handler/use-cases)

Building an agent, Agent Handler for Employees, or local development. A short comparison so you know which path applies before you click into a setup guide.