Skip to main content

Runlayer for Claude

Connect Claude to approved MCP servers through Runlayer, and import supported web and desktop chats through Anthropic’s Compliance API.

01

Connect

Give everyone in Claude the golden path to approved tools

Add one Runlayer Plugin endpoint so people find every tool they’re approved for, or connect a single server when you want a narrower set.

Runlayer Plugin

Unified endpoint

Add one remote MCP endpoint. Users discover and authorize the approved tools available to them without accumulating separate client configurations.

Direct connection

Per-connector endpoint

Give the client a dedicated Runlayer URL for one connector. Identity, policy, scanning, and audit still stay attached to every request.

Connection support

Remote MCP

Read the Claude setup guide
02

Observe

See how Claude is being used

Runlayer uses every channel Claude exposes. Each one is labeled, so you can tell real-time hooks from after-the-fact imports.

Supported

Anthropic inference hooks

With Claude Enterprise, Anthropic sends each governed Claude prompt to Runlayer before inference, with nothing installed on devices (Anthropic beta).

Read the guide
Monitor only

Compliance API import

Anthropic Compliance API imports Claude web and desktop chats for review.

Read the guide
Supported

Managed browser extension

Managed Chrome, Edge, and Firefox deployments can monitor supported web-chat activity and apply prompt-time controls where the surface permits it.

Read the guide
03

Enforce

Know exactly what Claude can block

Each row reflects Runlayer’s enforcement code for Claude: what it can stop before a call runs, block afterward, or redact in place.

ControlSupportWhat that means
Pre-tool denySupportedRunlayer can reject a Claude prompt before Anthropic runs it, which stops the whole request rather than one tool call.
Post-tool output blockSupportedBecause Runlayer checks the full prompt, tool results in it can be blocked before the model sees them.
Output rewriteUnsupportedAnthropic Inference hooks cannot return a rewritten prompt, so Runlayer cannot mask content in Claude.
Enforcement postureManaged device
Minimum supported versionNo minimum listed

Client-specific caveats

  • Runlayer checks the whole Claude prompt before Anthropic runs it, so a block stops that request rather than a single tool call.
04

Deploy at scale

Roll out Claude controls with the tools IT already uses

Admins can deploy the Runlayer CLI and endpoint package through MDM, then keep approved MCP configuration in sync automatically.

One signed package, settings managed centrally

Runlayer’s endpoint package installs on managed macOS and Windows devices. It discovers shadow AI, keeps hooks in place, captures sessions, and applies runtime controls. Approved MCP configuration re-syncs on a schedule for supported clients.

MCP servers

Connect Claude to the tools your teams use

Setup steps, tools, and controls for each MCP server in Claude.