autoresearch

Automate goal-directed iteration loops with mechanical success metrics in Claude Code.

1|Updated Apr 27, 2026
One-click install
npx skills add https://github.com/smaroc/best-skills-marketplace --skill autoresearch-smaroc
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: autoresearch
Source: https://github.com/smaroc/best-skills-marketplace/tree/main/plugins/best-skills/skills/autoresearch
Command: npx skills add https://github.com/smaroc/best-skills-marketplace --skill autoresearch-smaroc

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Autoresearch enables autonomous, goal-driven iteration that defines a measurable success criterion, runs iterative experiments, and learns from past results to continuously improve outcomes across tasks and domains.

Core Features & Use Cases

  • End-to-end autoresearch loop: define a goal, select a scope, verify progress with a mechanical metric, and decide keep/revert/rework automatically.
  • Interactive planning and gated execution: plan mode to craft executable configurations, and chain to related tools (debug, fix, ship, scenario, learn, predict, etc.).
  • Flexible workflow modes: unbounded loops for ongoing improvement or bounded iterations for CI-friendly gates; supports branching, memory, and automated rollback.
  • Rich domain coverage: used for software engineering, product discovery, and research documentation with built-in knowledge management and reports.

Quick Start

Start an unlimited autoresearch loop by invoking /autoresearch with a Goal and a Scope, then monitor progress and adjust as needed.

Frequently Asked Questions about autoresearch

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

FAQPage Schema
How do I automate goal-directed iteration for software engineering tasks?

Autonomous goal-directed iteration automates repetitive engineering tasks by defining a mechanical success metric, running bounded or unbounded loops, and chaining actions to continuously improve outcomes without manual intervention.

What is autonomous workflow planning and how does it handle errors?

Autonomous workflow planning uses a mechanical metric to verify progress during iterative loops, automatically deciding whether to keep, revert, or rework results based on dry-run validation before committing.

Can I use bounded iteration loops for CI-friendly automated testing?

Bounded iteration loops support CI-friendly gates by running a defined number of automated experiments, verifying progress with a mechanical metric, and halting based on your configured limits.

Do I need a dry-run verification command to automate product development workflows?

A dry-run verification command is required to validate progress and ensure mechanical success metrics are met before the system commits any automated product development or research results.

How does automated rollback work when iterative experiments fail?

Automated rollback triggers when dry-run verification fails the mechanical success metric, allowing the iteration loop to revert changes and rework the approach instead of committing flawed results.

What's the best way to start an unbounded iteration loop for research documentation?

Start an unbounded iteration loop by defining a research goal and scope, then monitor the automated experiments as they continuously learn from past results to improve documentation outcomes.