Skip to main content

​
Coding Agents

This document describes how to integrate AnyTask with AI coding agents including Claude Code, Cursor, Gemini CLI, and Codex CLI.

​
Overview

AnyTask provides slash commands that work with multiple AI coding agents, enabling you to:
  • Create tasks directly from your coding agent
  • Get task suggestions based on priority and dependencies
  • Work on tasks with full context loaded automatically
  • Track progress without leaving your development environment

​
Prerequisites

  1. AnyTask CLI installed
  2. Authentication configured (API key)
  3. Workspace initialized in your project

​
Quick Setup

​
Step 1: Install AnyTask CLI

See the Installation Guide for installation options.

​
Step 2: Authenticate

anyt login

​
Step 3: Initialize Workspace

cd /path/to/your/project
anyt init

​
Step 4: Install Agent Commands

# Interactive selection
anyt agent install

# Or install for specific agents
anyt agent install claude
anyt agent install cursor
anyt agent install gemini
anyt agent install codex

​
Supported Agents

AgentCommand FormatInstallation Location
Claude CodeMarkdown.claude/commands/
CursorMarkdown.cursorrules/
Gemini CLITOML.gemini/
Codex CLIYAML.codex/

​
Slash Commands

Once installed, use these slash commands in your coding agent:

​
/anyt:anyt-new {requirement}

Create a new task in AnyTask cloud. Example: 
User: /anyt:anyt-new Add user authentication with OAuth support

Agent: Creating task...
Created task DEV-42: Add user authentication with OAuth support
Priority: 1 | Status: todo

​
/anyt:anyt-next

Pick the next suggested task to work on. Example: 
User: /anyt:anyt-next

Agent: Top Task Recommendations:
1. DEV-42: Implement OAuth (Priority: 1)
   - All dependencies complete
   - Unblocks 2 other tasks

2. DEV-43: Add rate limiting (Priority: 0)
   - Blocked by: DEV-42

Would you like to start working on DEV-42?

​
/anyt:anyt-work {TASK_ID}

Continue working on a specific task. Example: 
User: /anyt:anyt-work DEV-42

Agent: Loading task DEV-42...

Task: Implement OAuth callback
Status: in_progress
Priority: 1

Description:
Add OAuth2.0 callback handler for Google authentication

Ready to continue implementation.

​
Example Workflows

​
Daily Development Workflow

User: /anyt:anyt-next
Agent: Top recommendation: DEV-42: Implement OAuth callback

User: Let's work on that
Agent: Starting work on DEV-42...

[You implement the feature]

User: I've finished the OAuth implementation
Agent: Great! Marking DEV-42 as done.
Next suggestion: DEV-43: Add rate limiting (now unblocked)

​
Creating Tasks from Conversations

User: We need to add rate limiting to the API

Agent: That sounds like important work. Would you like me to create a task for it?

User: Yes, please use /anyt:anyt-new

User: /anyt:anyt-new Add rate limiting to API endpoints

Agent: Created DEV-43: Add rate limiting to API endpoints
Priority: 1 | Status: todo

​
How It Works

  1. Slash commands execute anyt CLI commands in the background
  2. Task context is loaded from .anyt/tasks/{ID}/task.md
  3. Progress is synced back to AnyTask cloud automatically
  4. Active task is tracked in .anyt/active_task.json

​
File Locations

PathDescription
.anyt/anyt.jsonWorkspace configuration
.anyt/active_task.jsonCurrently active task
.anyt/tasks/Local task files

​
Troubleshooting

​
Commands not found

Ensure agent commands are installed by running anyt agent install.

​
Not in workspace

Initialize workspace in your project: 
anyt init

​
Authentication errors

Re-authenticate: 
anyt login
anyt health check

​
Next Steps