Skip to content

ai_assist

CLI tool to manage AI Assistant workspaces.

Usage:

$ ai_assist [OPTIONS] COMMAND [ARGS]...

Options:

  • --install-completion: Install completion for the current shell.
  • --show-completion: Show completion for the current shell, to copy it or customize the installation.
  • --help: Show this message and exit.

Commands:

  • init: Initialize a new AI Assistant workspace.
  • run: Start backend and frontend services.
  • workspace: Workspace management commands
  • global: Global resources management
  • knowledge: Knowledge base management
  • note: Customer note commands

ai_assist init

Initialize a new AI Assistant workspace.

Creates: - ~/.ai_assist/: Global directory for cross-workspace resources - prompts/: Shared prompt templates - agents/: Agent definitions - tools/: Shared tool definitions - models/: Model configurations - cache/: Shared cache

  • <workspace>/: Local workspace directory
  • data/chroma/: Vector database
  • prompts/: Local prompt templates
  • tools/: Local tool definitions
  • history/: Chat history
  • summaries/: Conversation summaries
  • notes/: Documents for RAG

Usage:

$ ai_assist init [OPTIONS] [PATH]

Arguments:

  • [PATH]: Path where to create the workspace. Defaults to current directory.

Options:

  • -n, --name TEXT: Workspace name. Defaults to directory name.
  • -f, --force: Overwrite existing workspace configuration.
  • --help: Show this message and exit.

ai_assist run

Start backend and frontend services.

By default, uses Docker Compose to start services with pre-built images. Use --dev flag to run services directly using uv and npm (development mode).

The command will auto-detect the workspace by looking for the workspace marker in the current directory or parent directories. You can also specify a workspace path.

Usage:

$ ai_assist run [OPTIONS] [WORKSPACE_PATH]

Arguments:

  • [WORKSPACE_PATH]: Path to workspace. Defaults to current directory (auto-detected).

Options:

  • -d, --dev: Run in development mode (uses uv and npm instead of Docker).
  • --help: Show this message and exit.

ai_assist workspace

Workspace management commands

Usage:

$ ai_assist workspace [OPTIONS] COMMAND [ARGS]...

Options:

  • --help: Show this message and exit.

Commands:

  • status: Show workspace status and directory summary.
  • clean: Clean workspace data (history, summaries,...
  • list: List all known workspaces.

ai_assist workspace status

Show workspace status and directory summary.

Usage:

$ ai_assist workspace status [OPTIONS] [PATH]

Arguments:

  • [PATH]: Path to workspace. Defaults to current directory.

Options:

  • --help: Show this message and exit.

ai_assist workspace clean

Clean workspace data (history, summaries, cache).

Usage:

$ ai_assist workspace clean [OPTIONS] [PATH]

Arguments:

  • [PATH]: Path to workspace. Defaults to current directory.

Options:

  • -y, --yes: Skip confirmation prompt.
  • --help: Show this message and exit.

ai_assist workspace list

List all known workspaces.

Usage:

$ ai_assist workspace list [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist global

Global resources management

Usage:

$ ai_assist global [OPTIONS] COMMAND [ARGS]...

Options:

  • --help: Show this message and exit.

Commands:

  • status: Show global AI Assist home status.
  • config: Display global configuration file.
  • prompts: List available global prompts.
  • agents: List available global agents.
  • tools: List available global tools.
  • tree: Show the global directory structure.

ai_assist global status

Show global AI Assist home status.

Usage:

$ ai_assist global status [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist global config

Display global configuration file.

Usage:

$ ai_assist global config [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist global prompts

List available global prompts.

Usage:

$ ai_assist global prompts [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist global agents

List available global agents.

Usage:

$ ai_assist global agents [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist global tools

List available global tools.

Usage:

$ ai_assist global tools [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist global tree

Show the global directory structure.

Usage:

$ ai_assist global tree [OPTIONS]

Options:

  • --help: Show this message and exit.

ai_assist knowledge

Knowledge base management

Usage:

$ ai_assist knowledge [OPTIONS] COMMAND [ARGS]...

Options:

  • --help: Show this message and exit.

Commands:

  • process: Process knowledge documents from a JSON...
  • stats: Show RAG vector store statistics.

ai_assist knowledge process

Process knowledge documents from a JSON specification file.

The JSON file should contain an array of document specifications:

[ {"document_type": "website", "uri": "https://...", "collection": "flink"}, {"document_type": "folder", "uri": "$HOME/docs", "collection": "python"} ]

Supported document_type values: website, folder, markdown

The 'collection' field is mapped to 'category' in the backend for filtering. Environment variables ($HOME) and ~ are expanded in URIs.

Usage:

$ ai_assist knowledge process [OPTIONS] JSON_FILE

Arguments:

  • JSON_FILE: Path to JSON file containing document specifications. [required]

Options:

  • -b, --backend-url TEXT: Backend API URL. Defaults to AI_ASSIST_BACKEND_URL env var or http://localhost:8000/api
  • -n, --dry-run: Validate JSON and show what would be processed without making API calls.
  • -f, --force: Force re-indexing even if content unchanged.
  • -v, --verbose: Show detailed output including errors.
  • --help: Show this message and exit.

ai_assist knowledge stats

Show RAG vector store statistics.

Usage:

$ ai_assist knowledge stats [OPTIONS]

Options:

  • -b, --backend-url TEXT: Backend API URL. Defaults to AI_ASSIST_BACKEND_URL env var or http://localhost:8000/api
  • --help: Show this message and exit.

ai_assist note

Customer note commands

Usage:

$ ai_assist note [OPTIONS] COMMAND [ARGS]...

Options:

  • --help: Show this message and exit.

Commands:

  • parse: Parse a customer note markdown file and...

ai_assist note parse

Parse a customer note markdown file and create organization, project, persons, and meeting refs.

The file should follow a structure like: H1 = org name, ## Team / ## <Org> = people,

Products/Context/Technology stack = context, ## Past steps/Next steps = steps,

Discovery call/Questions = meeting sections. Uses an LLM to extract structured

data and optionally creates entities via the backend API.

Usage:

$ ai_assist note parse [OPTIONS] FILE_PATH

Arguments:

  • FILE_PATH: Path to customer note markdown file (e.g. index.md). [required]

Options:

  • -b, --backend-url TEXT: Backend API URL. Defaults to AI_ASSIST_BACKEND_URL or http://localhost:8000/api
  • -n, --dry-run: Only parse and show extracted data; do not call the backend.
  • -v, --verbose: Show detailed output.
  • --help: Show this message and exit.