Claude

Bring Claude Code to governed tools and APIs.

Use Claude Code as an MCP client for scoped GitHub, package-backed, or imported tool access through ExecWarden.

Integration shape

The setup, credentials, governed actions, and evidence should be obvious.

Compatibility

Tested MCP registration and bearer-authenticated connectivity for Claude Code.

Agent loop

Claude Code owns the local or external agent loop.

Tools

ExecWarden exposes selected capabilities through MCP.

Credentials

The Claude client receives scoped MCP access, not broad underlying provider credentials.

Governed actions

Calls through ExecWarden can be scoped, approval-gated, logged, expired, and revoked.

Not implied

This is MCP connectivity and governed tool access, not a hosted Claude model session inside ExecWarden.

Start

Use Claude Code as an MCP client with a scoped Agent identity.

How it works today

How this boundary works in practice.

Claude Code owns the agent loop. ExecWarden supplies the scoped capability boundary for the tools it calls through MCP.

MCP client

Connect Claude through the same capability boundary.

Use an Agent identity to expose only the tools the task needs.

Approvals

Pause sensitive writes before they leave ExecWarden.

Ask mode can hold exact action arguments for review where the capability supports it.

Revocation

Stop future access without rotating every provider key.

Disable, expire, revoke, or rotate the Agent identity when the job is done.

Examples

Example tasks and outcomes.

1Claude Code uses a scoped MCP identity for repo work.
2A package-backed write pauses for approval.
3The activity record shows the tool call and decision path.

Get started during beta

Start with one bounded task and one boundary to review.

Use the app during beta, or open the docs and choose the first agent task you want to govern.

Connect an MCP clientRead MCP quickstart