What happens if the code fails the review phase?

The skill enters an automated fix loop, where it attempts to address the requested changes and re-verify the code, up to a maximum of 3 iterations by default.

How do I pick up work from a previous session?

You can use the --resume latest command to continue from your most recent session, or provide a specific session ID to return to a previous state.

How does smart parallel detection work?

The skill analyzes tasks to see if they edit the same files. If tasks are independent, it can run up to 3-5 workers simultaneously to significantly speed up implementation.

Can I prevent the skill from automatically committing code?

Yes, you can use the --no-commit flag in the command line or set auto_commit to false in your .claude-code-harness.config.yaml file.

Autonomous Task Implementation

Name: Autonomous Task Implementation
Author: Chachamaru127

byChachamaru127

•

160

Productivity & Workflow

Executes complex development tasks from Plans.md using smart parallel processing and a multi-step review loop.

About

The Work skill is a core component of the Claude Code development harness designed to transition from planning to implementation seamlessly. It automates the execution of tasks defined in a project's Plans.md file, utilizing intelligent parallel detection to run independent tasks concurrently for maximum efficiency. Beyond simple code generation, it incorporates a robust 'Plan-Work-Review' cycle, where code is implemented, verified, and reviewed through multiple perspectives to ensure high-quality output. With advanced session management for forking and resuming work, it provides a professional-grade autonomous workflow for building complex software features.

Key Features

Smart parallel execution with dependency and file-conflict detection
CI-ready non-interactive mode with customizable auto-commit logic
Automated implementation-fix cycles with configurable iteration limits
160 GitHub stars
Session management allowing for session forking and resuming
Multi-perspective review loop for automated quality assurance

Use Cases

Implementing multi-file features based on an architectural plan
Accelerating large-scale refactoring through concurrent task processing
Automating the build-verify-review cycle in local or CI environments

About

Key Features

Smart parallel execution with dependency and file-conflict detection
CI-ready non-interactive mode with customizable auto-commit logic
Automated implementation-fix cycles with configurable iteration limits
160 GitHub stars
Session management allowing for session forking and resuming
Multi-perspective review loop for automated quality assurance

Use Cases

Implementing multi-file features based on an architectural plan
Accelerating large-scale refactoring through concurrent task processing
Automating the build-verify-review cycle in local or CI environments