What signals increase the auto-continue confidence score?

Confidence increases when Claude detects positive signals such as successful builds, passing tests, completed milestones, and progress indicators.

What is the auto-continue skill for Claude Code?

It is a specialized capability that allows Claude to automatically proceed with sequential tasks by evaluating its own confidence score, significantly reducing the need for manual user approval during long workflows.

How do I adjust the automation confidence threshold?

You can customize the threshold by setting the INFERENCE_CONTINUEZ_THRESHOLD environment variable or by modifying the settings file in the plugin root.

Is it safe to use auto-continue for destructive commands?

No, the skill is designed with safety guards that ensure destructive or high-risk operations always prompt the user for confirmation, regardless of the confidence score.

Auto-Continue for Claude Code

Name: Auto-Continue for Claude Code
Author: BLERBZ

byBLERBZ

Productivity & Workflow

Automates Claude Code task progression using intelligent confidence scoring to minimize manual interruptions during extended development workflows.

About

The auto-continue skill enhances Claude Code by implementing a confidence-based decision engine that determines when to proceed with the next step of a task automatically. By evaluating factors like task completion milestones, passing tests, and build success against a user-defined threshold, it eliminates the need for manual 'yes' prompts during long-running coding sessions or Ralph Loop iterations. It prioritizes safety by requiring manual intervention for destructive or high-risk operations while providing a seamless, near-autonomous experience for standard development cycles and monitoring tasks.

Key Features

0 GitHub stars
Detailed decision logging for audit and transparency
User-definable confidence threshold via environment variables
Built-in safety guards for destructive or dangerous operations
Seamless integration with Ralph Loop iteration cycles
Intelligent confidence scoring for automated task progression

Use Cases

Monitoring and processing background tasks in autonomous agent environments
Running autonomous iteration loops for project refactoring or bug fixing
Extended coding sessions where manual confirmation of every step is inefficient

About

Key Features

0 GitHub stars
Detailed decision logging for audit and transparency
User-definable confidence threshold via environment variables
Built-in safety guards for destructive or dangerous operations
Seamless integration with Ralph Loop iteration cycles
Intelligent confidence scoring for automated task progression

Use Cases

Monitoring and processing background tasks in autonomous agent environments
Running autonomous iteration loops for project refactoring or bug fixing
Extended coding sessions where manual confirmation of every step is inefficient