// MCP TOOL REFERENCE

JISHI TOOLS

7 tools from the Jishi MCP Server, categorised by risk level.

READ TOOLS

download_icons Download icon nodes as SVG. Uses selection if nodeIds not provided. get_node_children Get all children nodes of a specific node (by name or ID). get_page_nodes Get all top-level nodes (frames/artboards) in the current page. Useful for discovering available nodes before using get_node_children or get_select... get_selection Get the current selection in the editor, including detailed properties and all children recursively. list_plugin_clients List connected JiShi plugin clients for multiplexing.

WRITE TOOLS

save_image Export a node as an image. If savePath is provided, saves to that path on the server. Otherwise triggers download in the browser. 3/5

EXECUTE TOOLS

execute_script Execute JavaScript code in the JiShi Design plugin context. Use 'jsDesign' global object. The code should return a value (it will be wrapped in a f... 4/5

// FAQ

How many tools does the Jishi MCP server have? +

The Jishi MCP server exposes 7 tools across 3 categories: Read, Write, Execute.

How do I enforce policies on Jishi tools? +

Use Intercept, the open-source MCP proxy. Write YAML rules for each tool — rate limits, argument validation, or deny rules — then run Intercept in front of the Jishi server.

What risk categories do Jishi tools fall into? +

Jishi tools are categorised as Read (5), Write (1), Execute (1). Each category has a recommended default policy.

Enforce policies on Jishi

Open source. One binary. Zero dependencies.

npx -y @policylayer/intercept

github.com/policylayer/intercept →