@tasksai/install

Official installer CLI for TasksAI MCP verticals.

Users normally run this through a product-specific GitHub manifest, for example:

npm exec --package=@tasksai/install --call 'tasksai-install lawtasksai --source https://github.com/TasksAI-Official/tasksai-mcp-wrappers/tree/main/verticals/lawtasksai'
npm exec --package=@tasksai/install --call 'tasksai-install realtor --source https://github.com/TasksAI-Official/tasksai-mcp-wrappers/tree/main/verticals/realtor'

In restricted agent environments, grant write access to the default TasksAI application data folder and the selected MCP client config. If the runtime must be installed elsewhere, pass an exact product install directory:

npm exec --package=@tasksai/install --call 'tasksai-install farmer --source https://github.com/TasksAI-Official/tasksai-mcp-wrappers/tree/main/verticals/farmer --install-dir /tmp/tasksai/farmer'
npm exec --package=@tasksai/install --call 'tasksai-install realtor --source https://github.com/TasksAI-Official/tasksai-mcp-wrappers/tree/main/verticals/realtor --install-dir /tmp/tasksai/realtor'

TASKSAI_INSTALL_DIR=/tmp/tasksai/farmer is equivalent to --install-dir.

The installer:

• verifies the official vertical manifest
• installs the local MCP runtime
• connects the user's TasksAI account through browser approval when available
• stores credentials outside MCP client config
• configures supported MCP clients
• checks required write access before changing local files
• prints customer-facing recovery instructions when an AI assistant is blocked from writing local config files
• supports Claude Desktop, Cursor, Windsurf, and Codex
• runs a local health check
• exposes a local save-document tool for finished skill outputs

TasksAI servers handle authentication, credits, catalog/search metadata, and licensed skill delivery. TasksAI does not process user task content.

Local document output

After a workflow is executed, the customer's AI assistant applies the expert framework locally. When the final deliverable is ready, the assistant can call the vertical-specific save tool, such as lawtasksai_save_document or farmertasksai_save_document, to create a downloadable file on the customer's machine.

By default, files are saved under:

~/Documents/TasksAI/<ProductName>/

Set TASKSAI_OUTPUT_DIR in the local runtime environment to use a different folder. The save tool supports .docx and Markdown output, sanitizes filenames, and avoids overwriting existing files unless explicitly told to overwrite. Generated document content is handled by the customer's AI framework and local MCP runtime; it is not sent to TasksAI websites or APIs.