wtbatch

Orchestrates parallel subagents in isolated git worktrees to execute batches of work items.

1|Updated Dec 18, 2009
One-click install
npx skills add https://github.com/thurn/dotfiles --skill wtbatch-thurn
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: wtbatch
Source: https://github.com/thurn/dotfiles/tree/main/.llms/skills/wtbatch
Command: npx skills add https://github.com/thurn/dotfiles --skill wtbatch-thurn

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Executing a mixed batch of bugs, features, and chores one at a time is slow, while running them together risks file conflicts, broken dependencies, and tangled commits. This Skill coordinates the whole batch: it triages items, groups them by dependency and file overlap, dispatches subagents into isolated git worktrees, and consolidates everything into a single review-and-promote decision. ## Core Features & Use Cases - Batch triage and grouping: Sizes each work item (S/M/L), detects dependencies and shared files, and groups items so parallel worktrees never conflict. - Parallel subagent execution: Creates fresh git worktrees per group, dispatches subagents with explicit model selection, commit contracts, and evidence requirements (tests, screenshots), capped at about four concurrent groups. - Consolidated promotion: Presents one combined review with inline screenshots and demo URLs, then serially rebases and fast-forwards approved groups onto master with a ledger for crash recovery. - Use Case: You have a backlog of eight unrelated fixes and small features. Invoke this Skill with the list, and it plans the groups, runs them concurrently in separate worktrees, and returns one review prompt where you pick which groups to merge. ## Quick Start Ask the AI to run the wtbatch skill on your list of tasks, for example: use wtbatch to handle these items — fix the login redirect bug, add dark mode to settings, and update the onboarding copy.

Frequently Asked Questions about wtbatch

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run multiple coding tasks in parallel with AI agents?▼

Group the tasks by dependency and file overlap, then dispatch one subagent per group into its own git worktree so they run concurrently without conflicts. This Skill automates that triage, dispatch, monitoring, and consolidation workflow.

How do git worktrees prevent conflicts between parallel agents?▼

Each agent works in a separate worktree on its own branch, so parallel edits never touch the same checkout. Items that share files are deliberately placed in the same group, and promotion happens serially via rebase and fast-forward merges onto master.

Can dependent tasks be split across parallel worktrees?▼

No. Dependent items must share a group so the later item builds on the earlier one's commits. If the chain is too large for one subagent's context, sequential subagents run in the same worktree, each building on the previous commits.

What happens if a subagent reports done without test evidence?▼

A DONE report without verification evidence, such as test output or screenshots for visual changes, is rejected and sent back. The planner requires commit hashes, verification results, and screenshot paths before accepting any item as complete.

When should I not use batch worktree orchestration?▼

Avoid it for a single small task, where one worktree or direct editing is simpler, or when items are too ambiguous to size. Ambiguous items should be clarified with the user before dispatching any subagents.