MCP server for Xcode integration, providing access to the Xcode IDE for building and running Apple platform projects
Server has 4 tools with basic descriptions and no input schemas visible for most. Tool names follow verb_noun convention (list_*, take_*), which is good. However, descriptions are minimal (10-50 chars), parameter documentation is absent, and output schemas are not documented. The migrate_to_drews_xcode_mcp tool is unusual, it's a migration helper, not a core capability. No evidence of error handling guidance, recovery paths, or structured output documentation. Strict argument validation is implemented (good defensive practice), but tool definitions themselves lack the depth needed for reliable LLM reasoning.
List all on-screen macOS application windows with their CGWindow IDs. These window IDs can be used to capture screenshots of a given window or app with `take_app_screenshot` or `take_window_screenshot`.
List all currently running macOS applications.
Update this MCP server's configuration from the legacy xcode-mcp-server package name to drews-xcode-mcp.
Take screenshots of all windows for an app (case-insensitive substring match). If the app has more than one window, screenshots will be taken for up to 5 of them.
No input schemas visible for 3 of 4 tools (list_mac_app_windows, list_running_mac_apps, migrate_to_drews_xcode_mcp). Only take_app_screenshot shows a schema with app_name parameter. Without schemas, LLMs cannot validate inputs or understand parameter constraints.
Descriptions are too short (10 - 50 chars). Rubric baseline is 194 chars average; A+ tools range 50 - 200 chars. Current descriptions lack WHEN to use, WHAT it returns, and any prerequisites. E.g., 'List all on-screen macOS application windows' doesn't explain why an LLM would call this or what to do with the result.
No output schemas documented. LLMs need to know what fields to expect from list_mac_app_windows (does it return CGWindow IDs, app names, coordinates?), list_running_mac_apps (process IDs, bundle identifiers?), and take_app_screenshot (file paths, base64 data, metadata?). Without this, agents cannot plan downstream calls or extract the right data.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | 2026-07-28+ | v2 |
take_app_screenshot parameter app_name lacks description. The schema shows type:string but no guidance on case sensitivity, partial vs full match, or what happens if multiple apps match. Description says 'case-insensitive substring match' but this is in the tool description, not the parameter description, LLMs may miss it.
No error handling guidance. What happens if list_mac_app_windows finds no windows? If take_app_screenshot can't find the app? If screenshot capture fails? Descriptions don't explain recovery paths or what the LLM should do next.