For AI agents: a documentation index is available at the root level at /llms.txt and /llms-full.txt. Append /llms.txt to any URL for a page-level index, or .md for the markdown version of any page.
System StatusStart using Credal
  • Getting Started
    • Introduction
    • Quickstart
    • Video: Agent Building Basics
    • FAQs
  • Platform
      • Overview
      • Create a new Server
      • Attaching connectors
      • Connecting Agents
      • Testing your Server
      • Publishing your Server
      • Advanced Usage
    • Bulk Analysis
LogoLogo
System StatusStart using Credal
On this page
  • When to use it
  • Human approval for tool calls
  • Required server instructions
  • What’s next
PlatformMCP Server Builder

Advanced Usage (Beta)

Was this page helpful?
Edit this page
Previous

Third-party Server Registration

Next
Built with

Each MCP server you publish has an advanced endpoint available alongside its standard URLs. The advanced endpoint connects to the same server but turns on additional capabilities. You’ll find its URL in the Publish tab under Generic MCP Clients — copy it into your MCP client the same way you would any other MCP URL.

advanced-endpoint

When to use it

Use the advanced endpoint when you want either of the following.

Human approval for tool calls

Actions can be configured to require user approval before they run. These actions only work through the advanced endpoint — on the standard endpoint, they’re unavailable.

human-approval-toggle

When the user invokes one of these actions, they’ll be asked to approve it before it executes:

  • If the MCP client supports in-product approval prompts (the MCP “elicitations” feature), the prompt appears directly in the client.

    mcp-human-approval

  • Otherwise, the user is asked to approve the action in Slack.

Required server instructions

The server can require the client to fetch a set of starting instructions at the beginning of every session before any other tools become available. This is useful when there’s context, guidance, or guardrails the agent must read before acting.

What’s next

More capabilities will be added to the advanced endpoint over time.