anyt 0.7.1

AI-native task management CLI for coding agents and human developers

AnyTask CLI

AI-native task management CLI for coding agents and human developers

AnyTask CLI is a command-line interface for managing tasks and projects, designed specifically for AI agents and developers working together. It provides Linear-style task management with agent-aware features, MCP (Model Context Protocol) integration, and powerful AI-assisted commands.

Features

• Linear-style Task Management: Human-readable task IDs (DEV-123), status workflows, priorities, and dependencies
• Secure API Key Authentication: Simple API key-based authentication with optional keyring storage
• Workspace & Project Organization: Isolated workspaces with role-based access
• Rich Terminal UI: Beautiful, color-coded task boards and visualizations
• Agent Integration: Native support for AI agents via API keys and MCP server
• Worker System: Automated task execution using declarative YAML workflows
• AI-Powered Commands: Task decomposition, smart suggestions, and context-aware assistance

Installation

Recommended: Install with uvx or pipx

For the best experience, install using uvx (recommended) or pipx:

# Install uv first (if not already installed)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Using uvx (recommended - no installation needed)
uvx anyt --help

# Or using pipx (install once, use everywhere)
pipx install anyt
anyt --help

Alternative: Install with pip

pip install anyt

Quick Start

1. Get an API Key

Visit https://anyt.dev/home/settings/api-keys to create an API key.

2. Set Up Authentication

Add to your shell profile for persistence:

# For zsh users (macOS default)
echo 'export ANYT_API_KEY=anyt_agent_xxxxxxxxxxxxx' >> ~/.zshrc
source ~/.zshrc

# For bash users
echo 'export ANYT_API_KEY=anyt_agent_xxxxxxxxxxxxx' >> ~/.bashrc
source ~/.bashrc

Or set temporarily for current session:

export ANYT_API_KEY=anyt_agent_xxxxxxxxxxxxx

3. Initialize Your Workspace

Run the interactive setup:

anyt init

The CLI will guide you through:

• Verifying your authentication
• Interactively selecting a workspace from your available workspaces
• Interactively choosing a project to work in with rich display
• Saving configuration to .anyt/anyt.json
• Viewing completion summary and next steps

For automation/CI/CD:

# Auto-select first workspace and project
anyt init --non-interactive

# Or specify explicitly
anyt init --workspace-id 123 --project-id 5

4. Start Using AnyTask

# Create a task
anyt task add "Implement user authentication" --status todo --priority 1

# View your board
anyt board

# Pick a task to work on
anyt task pick DEV-1

# Show active task
anyt active

# Mark as done
anyt task done

Core Commands

Initialization & Configuration

anyt init                         # Interactive workspace setup (recommended)
  --non-interactive, -y           # Auto-select first workspace/project
  --workspace-id <id>             # Specify workspace ID
  --project-id <id>               # Specify project ID
  --identifier <prefix>           # Workspace identifier (e.g., DEV)
  --dev                           # Use development API

Health Check

anyt health check                 # Check backend server status

Task Management

Create and update tasks:

anyt task add <title>             # Create new task
  --status <status>               # backlog, todo, active, done, cancelled
  --priority <-2 to 2>            # Priority level (default: 0)
  --owner <user-id>               # Assign to user or agent
  --labels <labels>               # Comma-separated labels
  --phase <phase>                 # Phase/milestone identifier

anyt task create <title>          # Create from template
  --template <name>               # Template to use
  --no-edit                       # Skip editor

anyt task edit <identifier>       # Update task fields
  --title <title>                 # New title
  --status <status>               # New status
  --priority <priority>           # New priority (-2 to 2)
  --labels <labels>               # Replace all labels

anyt task done <identifier>       # Mark task(s) as done
  --note <message>                # Add completion note

anyt task rm <identifier>         # Delete task
  --force                         # Skip confirmation

anyt task bulk-update <ids>       # Update multiple tasks
  --status <status>               # New status for all
  --priority <priority>           # New priority for all
  --yes                           # Skip confirmation

View tasks:

anyt task list                    # List all tasks
  --status <status>               # Filter by status
  --mine                          # Show only my tasks
  --assignee <user-id>            # Filter by assignee
  --labels <labels>               # Filter by labels
  --phase <phase>                 # Filter by phase
  --sort <field>                  # Sort by field
  --limit <n>                     # Max tasks to show

anyt task show <identifier>       # Show task details
  --show-metadata                 # Include workflow metadata

anyt board                        # View Kanban board
  --mine                          # Show only my tasks
  --group-by <field>              # Group by status, priority, owner, labels
  --compact                       # Compact display

anyt task suggest                 # Suggest next tasks
  --limit <n>                     # Number of suggestions
  --include-assigned              # Include assigned tasks

Task dependencies:

anyt task dep add <identifier>    # Add dependencies
  --on <task-ids>                 # Tasks this depends on

anyt task dep rm <identifier>     # Remove dependencies
  --on <task-ids>                 # Tasks to remove dependency on

anyt task dep list <identifier>   # List dependencies

anyt graph <identifier>           # Visualize dependencies
  --format <ascii|dot|json>       # Output format
  --depth <n>                     # Max dependency depth

Active task tracking:

anyt task pick <identifier>       # Pick task to work on
anyt active                       # Show current task

Comments:

anyt comment add <identifier>     # Add comment
  --message <text>                # Comment content

anyt comment list <identifier>    # List all comments

Sharing:

anyt task share <identifier>      # Generate shareable link
  --copy                          # Copy to clipboard

Visualization & Reporting

anyt summary                      # Workspace summary
  --period <today|weekly|monthly> # Summary period
  --phase <phase>                 # Filter by phase
  --format <text|markdown|json>   # Output format

anyt timeline <identifier>        # Task timeline
  --since <YYYY-MM-DD>            # Show events since date
  --last <24h|7d>                 # Show last N hours/days
  --compact                       # Compact format

Worker System

Automated task execution for AI agents:

anyt worker start                 # Start worker (interactive agent selection)
  --agent-id <id>                 # Agent identifier (or select interactively)
  --workspace <path>              # Workspace directory
  --workflow <name>               # Specific workflow to run
  --workflows <dir>               # Custom workflows directory
  --poll-interval <seconds>       # Polling interval
  --project-id <id>               # Project scope

anyt worker list-workflows        # List available workflows
anyt worker validate-workflow <file>  # Validate workflow definition

Built-in workflows:

• local_dev: Direct implementation on current repository
• dry_run: Testing workflow without external dependencies
• feature_dev: Streamlined feature development workflow
• feature_dev_with_check: Feature development with comprehensive validation

Secret management:

anyt worker secret set <name>     # Store secret
  --value <value>                 # Secret value (or prompt)

anyt worker secret get <name>     # Retrieve secret
  --show                          # Show value (default: masked)

anyt worker secret delete <name>  # Delete secret

anyt worker secret test <text>    # Test secret interpolation

Workflow artifacts & attempts:

anyt attempt list <task-id>       # List task attempts
anyt attempt show <attempt-id>    # Show attempt details

anyt artifact download <id>       # Download artifact
  --output <path>                 # Output file
  --stdout                        # Print to stdout

Usage Examples

Daily Workflow

# View your board
anyt board

# Pick a task
anyt task pick DEV-42

# Show active task
anyt active

# Update progress
anyt task edit DEV-42 --status active

# Add a comment
anyt comment add DEV-42 -m "Completed implementation"

# Complete the task
anyt task done DEV-42 --note "All tests passing"

Team Collaboration

# List all tasks
anyt task list

# Filter by status and phase
anyt task list --status active --phase T7

# View task details
anyt task show DEV-123

# Add dependencies
anyt task dep add DEV-124 --on DEV-123

# Visualize dependency graph
anyt graph DEV-124

# Share task link
anyt task share DEV-123 --copy

Workflow Automation

# Get task suggestions for agent
anyt task suggest --limit 5

# Start worker with interactive agent selection
export ANYT_API_KEY=anyt_agent_xxx
anyt worker start --workflow feature_dev

# Or start worker with specific agent and workflow
anyt worker start --agent-id agent-xxx --workflow feature_dev_with_check

# View workflow attempts
anyt attempt list DEV-42
anyt attempt show attempt-123

# Download artifacts
anyt artifact download artifact-456 --output result.json

Bulk Operations

# Update multiple tasks at once
anyt task bulk-update DEV-1,DEV-2,DEV-3 --status done --yes

# Mark multiple tasks as done
anyt task done DEV-4 DEV-5 DEV-6

MCP Server Integration

AnyTask includes a Model Context Protocol (MCP) server for seamless integration with Claude Code and other AI tools.

The MCP server is bundled with the CLI - no separate installation needed. Simply configure it in your Claude Code settings to access your tasks directly from AI conversations.

# Start the MCP server
anyt mcp start

Configure Claude Code by adding to your ~/.config/claude/claude_desktop_config.json:

{
  "mcpServers": {
    "anytask": {
      "command": "uvx",
      "args": ["anyt", "mcp", "start"],
      "env": {
        "ANYT_API_KEY": "anyt_agent_xxxxxxxxxxxxx"
      }
    }
  }
}

Configuration

Environment Variables

Authentication and API configuration via environment variables:

# Required: API key for authentication
export ANYT_API_KEY=anyt_agent_xxxxxxxxxxxxx

# Optional: Override API URL (defaults to https://api.anyt.dev)
export ANYT_API_URL=http://localhost:8000

# Optional: Override config directory (defaults to ~/.config/anyt)
export ANYT_CONFIG_DIR=/custom/path

Workspace Configuration

After running anyt init, workspace settings are stored in .anyt/anyt.json:

{
  "workspace_id": 123,
  "workspace_name": "Development",
  "workspace_identifier": "DEV",
  "api_url": "https://api.anyt.dev",
  "current_project_id": 456,
  "last_sync": "2025-11-14T10:30:00Z"
}

Note: The workspace config file does not store API keys. All authentication is handled via the ANYT_API_KEY environment variable.

Active Task Tracking

The currently picked task is stored in .anyt/active_task.json:

{
  "identifier": "DEV-42",
  "title": "Implement OAuth callback",
  "picked_at": "2025-11-06T10:30:00Z"
}

Documentation

User Guides

• CLI Usage Guide - Complete CLI usage documentation
• CLI Reference - Auto-generated command reference
• Worker Guide - Worker system setup and usage

Developer Documentation

For contributors and internal development:

• CLAUDE.md - Project overview and development guide
• DEVELOPER_GUIDE.md - Development setup and workflow
• docs/ - Complete documentation index

Support

For issues and questions, contact: [e-mail]

License

Proprietary - AnyTransformer Inc.