Understanding MCP (Model Context Protocol)
What is MCP?
Model Context Protocol (MCP) is an open standard that lets external AI applications connect to tools and data sources in a structured way. Instead of each AI client building custom integrations, MCP provides a common protocol for AI clients to discover tools, read resources, and call functions.
Chase Agents as an MCP server
Chase Agents exposes its own MCP server that external AI clients — such as Claude Desktop, Claude Code, or ChatGPT — can connect to. Once connected, those external clients can use the full Chase Agents execution engine: building automations, running multi-step workflows, calling connected APIs, executing Python code, generating documents, and managing automation schedules.
Available MCP tools
The Chase Agents MCP server exposes tools across several categories:
- execute_steps — run one or more workflow steps with full support for API calls, Python code, MCP tool calls, and control flow
- http_request — make a single HTTP request to a configured API connection
- run_python_code — execute Python in an isolated sandbox
- call_mcp_tool — call a tool on any connected MCP server
- create_automation, view_automation, manage_automation — full automation lifecycle management
- test_automation, run_automation, get_automation_runs — testing, running, and debugging
- manage_schedules — create, update, pause, and delete cron or interval schedules
- create_agent — create a new AI agent in the workspace
- manage_knowledge — store and retrieve notes about API connections
- list_connections — list all API and MCP server connections in the workspace
Guide resources via MCP
The MCP server also exposes detailed guide resources. External clients can use the read_resource tool to access documentation on step types, data flow patterns, control flow, automation building, and critical execution rules. This means external AI assistants can look up the correct patterns rather than guessing.
Connecting an external client
External AI clients connect to the Chase Agents MCP server using a workspace API key. Once connected, all operations run in the context of that workspace, with access to its connections, automations, and agents. See the MCP server documentation for your specific client (Claude Desktop, Claude Code, etc.) to configure the connection.