Skip to content
Get packs

Autoresearch Project Manager

Runs multi-step autonomous AI research projects: experiments, literature, findings and progress reports

Source Download skill
Repository
Orchestra-Research/AI-Research-SKILLs
Stars
14,195
Licence
MIT
Files
8 files

What it does

This skill turns an AI coding agent into a research project manager. Once installed, when you start a research question, the agent sets up a structured project folder with state, logs and findings files, searches literature, forms hypotheses, runs experiments in repeated cycles, periodically reflects on results, and produces HTML progress presentations so you can check on it. It is designed to keep working without asking for permission at each step, and it hands domain-specific work to other skills.

Great for

  • ML or AI researchers who want an agent to run long experiment campaigns while they are away
  • Students exploring a research question who want organized notes, experiment logs and a draft paper trail
  • Developers who already use Claude Code or similar agents and want a structured workflow for multi-hypothesis projects

Skip it if

  • Quick one-off questions or small coding tasks, since the workflow adds a lot of setup and overhead
  • Projects where experiments touch production systems, paid cloud compute or real user data, because the skill tells the agent to act without confirmation and it may run code or spend resources unattended
  • Anyone expecting verified scientific results: the agent's own synthesis and literature search can be wrong and need human review

Try asking

  • “Start an autoresearch project to test whether a smaller model keeps accuracy after quantization, and keep logging results”
  • “Resume my existing research project from research-state.yaml and continue the next experiment cycle”
  • “Review my research plan, set up the workspace, and run the inner experiment loop with progress presentations”
  • “Summarize findings so far in findings.md and propose new hypotheses from the results”
Step-by-step guide: how to use Autoresearch Project ManagerInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 8 original files, the MIT licence and an attribution note.

    0-autoresearch-skill.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip 0-autoresearch-skill.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip 0-autoresearch-skill.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Set a clear research question and a measurable optimization target before starting so the inner loop has a useful goal
  • Check the to_human/ presentations regularly and redirect the agent, since it is instructed not to ask for permission
  • Keep large checkpoints outside the project folder and review the workspace for any credentials or data the agent may write there

What's inside

  • references/agent-continuity.md
  • references/progress-reporting.md
  • references/skill-routing.md
  • SKILL.md
  • templates/findings.md
  • templates/progress-presentation.html
  • templates/research-log.md
  • templates/research-state.yaml

Redistributed unmodified from Orchestra-Research/AI-Research-SKILLs under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.