Skip to content
  1. Home
  2. Agent skills
  3. Autoresearch AI Plugin

Autoresearch AI Plugin

Autonomous edit, measure, keep-or-revert loop to optimise any measurable metric

What it is

A plugin with two skills: a general experiment loop that edits code, commits, runs a benchmark and keeps only improvements, logging state in a JSONL file that survives context resets; and an ML variant with GPU templates for LLM training. It includes noise-aware confidence scoring and stopping rules.

Good for

  • Reduce test suite runtime automatically
  • Shrink bundle size by iteration
  • Tune an LLM training run overnight

Install or run

/plugin marketplace add proyecto26/autoresearch-ai-plugin

Command as given in the project README. Read the repository before you run anything on your machine.

What you need

Claude Code; a runnable benchmark for the target metric

Worth knowingUnattended loops can burn many tokens; set iteration limits.

Sources

The summary is ours, written from the README. Spotted an error? See how we work.

More for data & research

See the topic →
  • arXiv Search

    NousResearch/hermes-agent

    Search arXiv papers by keyword, author, category or ID

    SkillGitHub stars: 263k
  • Blocked Page Recovery

    NousResearch/hermes-agent

    Recover content when a fetch hits 403, a paywall or a bot wall

    SkillGitHub stars: 263k
  • Grounded Citations

    NousResearch/hermes-agent

    Numbered inline citations backed by a URL ledger script

    SkillGitHub stars: 263k
  • xlsx

    anthropics/skills

    Create, edit and clean Excel and CSV spreadsheets

    SkillGitHub stars: 194k