MCP server

Read from the package source without running it. The installed server may list more.

• check_task_status

  Check the status of a running task execution.

• codmir_click

  Click at screen coordinates (x, y). Use after taking a screenshot to identify the target location.

• codmir_connect

  Establish a real-time agent session with the Codmir desktop app. This announces the AI provider to the desktop UI, which shows a colored border glow indicating which agent is controlling the screen. Call this at the start of any desktop control session.

• codmir_disconnect

  End the agent session with the Codmir desktop app. Removes the border glow indicator.

• codmir_double_click

  Double-click at screen coordinates (x, y).

• codmir_fetch_page

  Fetch a web page through the desktop agent. Returns HTTP status, headers, page title, all links, and the HTML body (truncated at 50KB). Use this for web app testing, crawling routes, and analyzing page content without needing screenshots.

• codmir_key

  Press a keyboard key or key combination. Use macOS key codes. Common: Return=36, Tab=48, Escape=53, Delete=51, Space=49, Up=126, Down=125, Left=123, Right=124. Flags: Shift=0x20000, Control=0x40000, Option=0x80000, Command=0x100000.

• codmir_list_directory

  List files and directories at a path on the desktop machine. Returns file names, sizes, and types. Useful for understanding project structure, finding config files, or mapping a codebase. Skips node_modules, .git, dist, etc.

• codmir_mouse_move

  Move the mouse cursor to screen coordinates (x, y) without clicking.

• codmir_right_click

  Right-click at screen coordinates (x, y) to open context menus.

• codmir_scan_routes

  Scan multiple routes on a web app and return status codes for each. Useful for finding 404s, broken pages, redirect chains, and 500 errors. Returns a summary (counts of 200/404/3xx/5xx) and per-route results.

• codmir_screen_info

  Get screen dimensions and the name of the frontmost application. Call this first to understand the display layout.

• codmir_screenshot

  Capture a screenshot of the entire screen. Returns a base64-encoded PNG image. Use this to see what is currently displayed.

• codmir_scroll

  Scroll at the given screen coordinates. Negative delta scrolls down, positive scrolls up.

• codmir_shellwrite action

  Run a shell command on the machine where Codmir desktop is running. Returns stdout, stderr, and exit code.

• codmir_status

  Check if the Codmir desktop agent is running and connected. Returns status, port, platform, and architecture.

• codmir_type

  Type text at the current cursor position. Use codmir_click first to focus the target input field.

• create_organizationwrite action

  Create a new organization/workspace for the user. Required before creating projects.

• create_projectwrite action

  Create a new project within an organization. Required before creating tickets.

• create_ticketwrite action

  Create a new ticket in Codmir. Use this to capture bugs, features, or tasks from the conversation.

• execute_taskwrite action

  Execute an AI-powered code task (bug fix, feature implementation, refactor). The task will be queued and executed by a Codmir agent.

• generate_sdk_code

  Generate SDK code for a specific use case. Returns ready-to-use TypeScript code.

• get_error_context

  Get enriched error context for AI debugging. Returns the full error report plus suggested files to investigate (extracted from stack frames), a human-readable summary, and similar resolved issues. Ideal for understanding and fixing production errors.

• get_project_context

  Get project overview including tech stack, recent activity, and team info.

• get_sdk_reference

  Get Codmir SDK API reference for a specific topic. Use this to learn how to use the SDK for tickets, projects, organizations, chat, or agent tasks.

• get_sprint_tasks

  Get tasks for the current or specified sprint. Returns all tasks with their status, assignee, and ticket details.

• get_workspace_status

  Check if user has organization and projects set up. Call this first if other tools fail.

• list_project_tasks

  List all tasks for a project.

• list_recent_errors

  List recent production errors from the Overseer error store. Returns the most recent errors with their refId, message, severity, route, and timestamp. Useful for getting an overview of current production issues.

• list_todo

  Get the to-do list for a project \u2014 tickets filtered as actionable items (open/in_progress), sorted by priority. Excludes tickets with auto-execution workflow enabled. Use this for an overview of what needs to be done.

• lookup_error

  Look up a production error report by any reference: refId (err_xxx), Next.js digest (3681244270), reportId (rpt_xxx), fingerprint (fp_xxx), or issueId (ISS-xxx). Returns the full stored error report with stack trace, breadcrumbs, and metadata.

• store_conversation

  Save this conversation as project documentation. Use to preserve important context, solutions, or decisions.

• update_todo_statuswrite action

  Update the status of a to-do item (ticket). Use to mark items as done, in progress, or reopen them.

• Code scan10 source files scanned25/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitylocal package, no credentials required12/15
• Maintenancerepository not readable: repo not found3/15
• Maintainer identityno repository or website to verify2/10

MCP: common questions

Is MCP server safe?

With care: it is graded C, so read the findings first (65/100). Read the MCP safety report

How do I install MCP?

It runs on your machine. Copy the Claude Code, Claude Desktop or Cursor config from the install section.

Does MCP need an API key?

Not as far as the registry entry and our scan can tell: no credentials are declared or required.

Is MCP maintained?

The latest release is v0.2.3.

What can I use instead of MCP?

Servers from other publishers that do the same job: Hunch MCP server, MCP server and Tailscale MCP server. Compare all MCP alternatives.