batch-tasks

Parse TODO.md steps, generate implementation plans, and execute them via claude -p.

8|5|Updated Feb 18, 2026
One-click install
npx skills add https://github.com/shenxingy/Clade --skill batch-tasks-shenxingy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: batch-tasks
Source: https://github.com/shenxingy/Clade/tree/main/configs/skills/batch-tasks
Command: npx skills add https://github.com/shenxingy/Clade --skill batch-tasks-shenxingy

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

batch-tasks removes the overhead of manually translating TODO checklist items into implementation work by generating detailed execution plans in advance and then running the selected steps unattended.

Core Features & Use Cases

  • TODO.md step parsing: Selects specific Step N sections and converts unchecked - [ ] items (with indented sub-details) into executable task descriptions.
  • Pre-execution planning in-context: Researches and reads relevant code before execution so claude -p sessions waste less time rediscovering the same files.
  • Risk control and resiliency: Scores each task for readiness, skips low-scoring ones with GitHub Issues, rolls back on failure, retries configurable attempts, and logs outcomes to PROGRESS.md.
  • Serial or parallel execution: Runs tasks sequentially or in parallel using git worktrees with conflict detection/merge handling.
  • Post-run workflow: Produces logs and summary so you can then commit and review behavior anchors.

Quick Start

Use batch-tasks to run the first incomplete TODO step by telling your AI: “/batch-tasks”.

Frequently Asked Questions about batch-tasks

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

FAQPage Schema
How do I automate executing TODO.md steps for a batch refactor unattended?▼

You can automate executing TODO.md steps by parsing unchecked items into executable task descriptions, generating pre-execution implementation plans, and then running them unattended via claude -p in fresh sessions.

Can I run autonomous coding tasks in parallel using git worktrees?▼

Yes, autonomous coding tasks can run in parallel using git worktrees, which isolates changes per task and includes built-in conflict detection and merge handling to safely combine the results.

How does autonomous coding handle risky or underspecified tasks in a TODO.md checklist?▼

Autonomous coding handles risky tasks by scoring each item for readiness during planning, skipping low-scoring ones by generating GitHub Issues, and applying automatic rollback on execution failure.

What is the best way to retry failed batch coding tasks and log the outcomes?▼

The best way to retry failed batch coding tasks is by configuring retry attempts during execution, logging outcomes to PROGRESS.md, and rolling back the repository state to ensure a clean baseline.

Do I need to manually specify files before running multi-step coding checklists unattended?▼

No, you do not need to manually specify files because the tool performs in-context code planning before execution, researching and reading relevant code to target specific files and lines automatically.

When should I not use unattended task planning for feature additions?▼

You should not use unattended task planning when tasks are highly underspecified or risky, as the system will skip low-scoring items and generate GitHub Issues instead of attempting execution blindly.