smithery.ai

taskyou

Install and manage TaskYou — a personal task management system with Kanban board, background AI execution, and git worktree isolation. Guides installation if not present, then orchestrates tasks via CLI.

First seen Mar 20, 2026

Installation

$ npx skills add https://smithery.ai

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

Also in this package

Other skills from smithery.ai · top by installs.

npx skills add https://smithery.ai

Browse all from smithery.ai

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Declared
Cursor Not declared
Codex Declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Declared
Cline Not declared
OpenCode Not declared

Skill metadata

Parsed from SKILL.md frontmatter.

Declared agents claude-code codex gemini

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 7,829 B
  • docs SUMMARY.md 158 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 1 installs

SKILL.md

TaskYou — Install & Manage

You are an autonomous orchestrator for TaskYou, a personal task management system with a Kanban board, background AI execution, and git worktree isolation.

First: Check Installation

Before running any commands, verify TaskYou is installed:

which ty || echo "NOT_INSTALLED"

If NOT installed, guide the user through installation:

Quick install (recommended):

curl -fsSL taskyou.dev/install.sh | bash

This auto-detects OS/arch (macOS and Linux, amd64 and arm64) and installs ty to ~/.local/bin/.

Options:

  • --no-ssh-server — Skip installing taskd (the SSH daemon for remote access)
  • Set INSTALL_DIR=/custom/path to change install location

After install, verify:

ty --version

If ~/.local/bin is not in PATH, tell the user to add it:

export PATH="$HOME/.local/bin:$PATH"

Upgrading an existing install:

ty upgrade

If already installed, proceed to task management below.


CLI Reference

Use ty (short) or taskyou (full) — both work identically.

Action Command
See all tasks ty board --json
List by status ty list --status <status> --json
View task details ty show <id> --json --logs
Create task ty create "title" --body "description"
Create a workflow ty pipeline "goal" --project <name> (plan → code → parallel review → collect)
List / author workflows ty pipeline --list · ty pipeline new "<describe it>" · ty pipeline edit
Execute task ty execute <id>
Retry with feedback ty retry <id> --feedback "..."
Change status ty status <id> <status>
Pin/prioritize ty pin <id>
Close/complete ty close <id>
Delete ty delete <id>
See executor output ty output <id>
Send input to executor ty input <id> "message"
Confirm prompt ty input <id> --enter

Statuses: backlog, queued, processing, blocked, done, archived

Core Workflow

1. Survey the Board

Always start by understanding the current state:

ty board --json | jq

This returns all columns and their tasks. Use it to:

  • See what's in progress
  • Find blocked tasks needing input
  • Identify the next priority in backlog

2. Execute Tasks

Queue a task for execution:

ty execute <id>

The executor (Claude Code, Codex, Gemini, or Grok) runs in an isolated git worktree. Monitor progress:

ty show <id> --json --logs

3. Handle Blocked Tasks

When tasks are blocked, they need user input. Check why:

ty show <id> --json --logs

Then retry with feedback:

ty retry <id> --feedback "Here's the clarification you need..."

4. Direct Executor Interaction

For running/blocked tasks, you can interact directly with the executor:

# See what the executor is doing/asking
ty output <id>              # Capture recent terminal output
ty output <id> --lines 100  # More history

# Send input directly to the executor's terminal
ty input <id> "yes"         # Send text and press Enter
ty input <id> --enter       # Just press Enter (confirm a prompt)
ty input <id> --key Down --enter  # Press Down arrow then Enter

This is useful when:

  • The executor is waiting for permission (just send --enter)
  • You need to respond to a TUI prompt (use --key for navigation)
  • You want to see what's happening without attaching to tmux

5. Create New Tasks

ty create "Implement feature X" --body "Detailed description here..."

Optional flags:

  • --project <name> — Associate with a project
  • --type <type> — Task type (bug, feature, etc.)
  • --dangerous — Skip permission prompts in executor

6. Manage Priority

Pin important tasks to the top:

ty pin <id>           # Pin
ty pin <id> --unpin   # Unpin
ty pin <id> --toggle  # Toggle

7. Change Status

Move tasks between columns:

ty status <id> backlog     # Move to backlog
ty status <id> queued      # Queue for execution
ty status <id> done        # Mark complete

8. Close and Cleanup

ty close <id>    # Mark as done
ty delete <id>   # Permanently remove

JSON Output

All commands support --json for machine-readable output:

ty board --json              # Full board snapshot
ty list --status blocked --json  # Filtered list
ty show 42 --json --logs     # Task with execution logs

Pipe to jq for processing:

ty board --json | jq '.columns.backlog.tasks[:3]'  # First 3 backlog items
ty list --json | jq '.[] | select(.pinned == true)'  # Pinned tasks only

Orchestration Patterns

Auto-Execute Queued Tasks

ty list --status queued --json | jq -r '.[0].id' | xargs -I{} ty execute {}

Continuous Monitoring

while true; do
  ty board --json | jq -r '.columns.blocked.tasks[].id' | while read id; do
    echo "Task $id is blocked - needs attention"
  done
  sleep 30
done

Task Lifecycle

backlog → queued → processing → done
                 ↘ blocked (needs input)
  • backlog: Created but not started
  • queued: Waiting for executor pickup
  • processing: Actively being worked on
  • blocked: Waiting for user input/clarification
  • done: Completed successfully

MCP Tools (When Running Inside TaskYou)

If you're running as the task executor inside a TaskYou worktree, the taskyou MCP server is auto-registered and these tools are auto-approved (no permission prompts). Use them instead of the CLI — the CLI is for orchestrators outside the worktree, not the executor inside it.

Completion signaling (REQUIRED — nothing else watches for completion):

  • taskyou_complete — Move your task to done with a summary. Call this

when you finish. Without it, the task stays in processing/blocked and the orchestrator has to close it by hand.

  • taskyouneedsinput — Move your task to blocked with a question for

the user. Use this instead of prompting in the terminal.

Project context (call FIRST before exploring):

  • taskyougetproject_context — Read cached codebase summary from a prior

task. If present, skip exploration.

  • taskyousetproject_context — Save your exploration summary so future

tasks in this project can skip it.

Task introspection:

  • taskyoushowtask — Get details of any task in the current project.
  • taskyoucreatetask — Create a follow-up task.
  • taskyoucreatepipeline — Spin up a multi-step workflow (plan → code →

parallel review → collect) for a goal, when it warrants more than one task.

  • taskyoulisttasks — See other active tasks in this project.

Best Practices

  1. Always check the board first — Understand context before acting
  2. Use JSON output — Structured data is easier to process and reason about
  3. Provide meaningful feedback — When retrying blocked tasks, give clear context
  4. Monitor execution — Use ty show <id> --logs to track progress
  5. Respect priorities — Check pinned tasks first
  6. Create focused tasks — One clear objective per task works best

Troubleshooting

Task stuck in processing?

ty show <id> --logs
ty output <id>  # See live terminal output

No tasks executing?

ty daemon status
ty daemon restart  # If needed

Can't find a task?

ty list --status done --json | jq '.[] | select(.title | contains("keyword"))'