autoresearch

Run a Modify-Verify-Keep/Discard loop across six operational modes.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/wendeus0/AI-dotfiles --skill autoresearch-wendeus0
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: autoresearch
Source: https://github.com/wendeus0/AI-dotfiles/tree/main/core/skills/autoresearch
Command: npx skills add https://github.com/wendeus0/AI-dotfiles --skill autoresearch-wendeus0

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Autonomous iterative improvement loops enable AI systems to continuously refine an artifact by cycling through Modify → Verify → Keep/Discard → Repeat, reducing manual intervention and accelerating learning.

Core Features & Use Cases

  • Six operational modes: loop, plan, debug, fix, security, ship, each tailoring how goals are translated into actions and gates.
  • Automatic loading of foundational references (core principles, runtime invariants) to guide decisions and enforce safety.
  • Atomic, single-change-per-iteration workflow with explicit guard and holdout checks to ensure measurable progress and rollback if needed.

Quick Start

Start an autoresearch run in loop mode to iteratively improve your target artifact until a stopping guard is reached.

Frequently Asked Questions about autoresearch

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

FAQPage Schema
How do I set up autonomous iterative improvement loops for AI tasks?

Autonomous iterative improvement loops cycle through Modify → Verify → Keep/Discard steps to refine artifacts. This process reduces manual intervention by running structured gates that enforce mechanical verification and explicit exit states per iteration.

What is the best way to automate debugging and fixing code with continuous verification?

Automating continuous verification uses debug and fix modes to translate goals into atomic changes with guard checks. Each single-change iteration is mechanically verified and rolled back if it fails, ensuring measurable progress without manual intervention.

How does the modify verify keep discard loop handle runtime invariants?

The modify verify keep discard loop enforces runtime invariants by automatically loading foundational references and core principles. This guides decisions and safety checks, ensuring explicit holdout validation and atomic single-change progress per iteration.

Can I use autonomous improvement workflows for security and shipping gates?

Autonomous improvement workflows support security and ship modes to gate automation safely. These modes translate vague goals into measurable configurations, applying runtime invariants and explicit exit states to validate security checks and shipping readiness.

Do I need predefined configurations to start an autonomous automation loop?

You do not need fully predefined configurations to start. The automation loop can take vague goals and translate them into measurable configurations and gates, using loaded core principles to guide the iterative refinement process.

When should I avoid using atomic single-change iterations for task automation?

You should avoid atomic single-change iterations when your task cannot be broken into mechanically verifiable steps. The loop requires explicit exit states and measurable progress, making it unsuitable for subjective artifacts lacking automated verification gates.