MCP Server

The Core implements a Model Context Protocol (MCP) server so AI assistants can query and control Linknesis.

Protocol

The MCP endpoint speaks JSON-RPC 2.0 over HTTP at POST /api/mcp. Supported methods include initialize, notifications/initialized, tools/list, and tools/call.

Available tools

list_probes, get_probe, update_probe, delete_probe, restore_probe
list_tests, get_test, start_test, stop_test, delete_test
list_test_results, list_raw_test_results, list_webhook_logs

Security

MCP uses its own bearer token and the same action-based permissions as the REST API. Write tools validate inputs to prevent accidental data loss.

Testing

The dedicated /mcp-conf-test page lets you exercise the MCP server through any configured AI provider and inspect the full request/response chain.