Back to directory

NanoResearch

An end-to-end autonomous research agent that carries a research idea through literature search, experiment design, real GPU training, result analysis, and a complete LaTeX paper with figures drawn from actual experiment data.

NanoResearch

NanoResearch is an autonomous research pipeline that runs actual computational experiments rather than only generating code: it submits training jobs to a GPU cluster (local or Slurm), collects the real results, and produces a full LaTeX paper whose figures and tables come from those runs.

The pipeline moves through stages: ideation (literature search across OpenAlex and Semantic Scholar, idea generation), planning (turning an idea into an experiment blueprint with datasets, baselines, metrics, and ablations), setup, coding, execution, analysis of training logs, figure generation, writing, and review. Work can resume from any stage after an interruption.

It provides both a full-screen terminal UI and a plain streaming-log CLI, integrates with Claude Code, and includes a Feishu bot entry point.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.