Top 200 ReposEXPLORER
Browse agents, categories and pages
Back to results
Reported by sourceThe source says this happened. Top 200 Repos has not reproduced it.

Answer technical questions by letting async agents run experiments

Simon Willison sends research questions to cloud coding agents such as Claude Code for web and Codex Cloud, which write and run experiments in a dedicated repo and open a PR.

The problem

How it works

He set up a public simonw/research repository with full network access and nothing sensitive in it, and launched two or three research tasks a day. Examples include benchmarking Python Markdown libraries, getting a C-extension library working in Pyodide, and testing scikit-learn tag suggestions for older blog posts. A GitHub workflow summarizes each project in the README.

Results and limitations

He reports 13 projects in about two weeks; the Markdown benchmark found cmarkgfm far faster than the six alternatives the agent located, and the Pyodide task produced a working 88.4KB wheel after two nudges.

Before you try it

He calls the reports essentially unreviewed and keeps them in a separate repo; agents can show something works but cannot prove it is impossible, and one task skipped a requested visualization. Open network access is only acceptable because the repo is disposable.

Tools and repositories

Tools
Claude Code for web · Codex Cloud · GitHub Actions · GitHub Models
Prerequisites
Needs access to a cloud coding agent and a GitHub repository dedicated to throwaway research.

Try it yourself: Send a code research question to an async agent

Fill the fields to adapt the prompt. Empty fields stay as visible [placeholders].

Preview
Research question: [Research question]. Work only inside the folder [Project folder] in this throwaway research repository. Write small scripts that test the question directly, run them, and save raw results to files rather than describing them from memory. If you install packages, list them with versions. Produce a README.md with the question, method, a results table, what you could not get working and why, and suggested next steps. Do not access private repositories, use credentials, or change files outside [Project folder]. Open a pull request titled [Pull request title] for my review instead of merging.

Use a separate repository with no secrets, because open network access exposes the agent to prompt injection.

Related videos

About anthropics/claude-code

About openai/codex

About anthropics/claude-code

Independent creators. Listing a video implies no endorsement, sponsorship or partnership.

Sources

  1. Code research projects with async coding agents like Claude Code and Codex (opens in a new tab)

    First-person account · simonwillison.net · Published 6 Nov 2025 · Source checked 9 Nov 2026, 22:28 UTC

    What this source supports
    • Uses Claude Code for web, Codex Cloud and others for code research
    • Dedicated simonw/research repo with 13 projects in two weeks
    • Markdown library benchmark and an 88.4KB Pyodide wheel
    • Author describes reports as largely unreviewed
Report a problem

Reported by source

Let a coding agent run overnight ML training experiments

Point a coding agent at autoresearch so it edits a small GPT training script, runs short timed experiments and keeps what improves.

github.comkarpathy/autoresearch, anthropics/claude-code, openai/codex
Reported by source

Run unattended ML training experiments with autoresearch

Red Hat's Roy Belio containerized Karpathy's autoresearch on OpenShift AI with H100 GPUs and let Claude Code run 198 training experiments over 24 hours.

developers.redhat.com · 7 Apr 2026karpathy/autoresearch, anthropics/claude-code
Planned, not done

Build a formula-driven Excel model with the Claude xlsx skill

Install Anthropic's xlsx skill in Claude Code so the agent writes live formulas, recalculates the workbook with LibreOffice and checks it for formula errors.

github.comanthropics/skills, anthropics/claude-code
Reported by source

Build an on-call investigation agent from your own alert history

Goutham Veeramachaneni connected Claude Code to Grafana's gcx CLI, his code and runbooks, then trained its skill on two weeks of past pages for his services.

gouthamve.dev · 21 Apr 2026anthropics/claude-code
Reported by source

Get a new OCR model running on unfamiliar GPU hardware with Claude Code in a sandbox

Simon Willison gave Claude Code full permissions inside a Docker container on an NVIDIA Spark and had it get DeepSeek-OCR working; it took four prompts and under 40 minutes, with one nudge from him.

simonwillison.net · 20 Oct 2025anthropics/claude-code, pytorch/pytorch, huggingface/transformers
Reported by source

Port a Python library to JavaScript with a coding agent

Simon Willison had Codex CLI translate the JustHTML parser into a dependency-free JavaScript library, steering it with about eight prompts over one afternoon.

simonwillison.net · 15 Dec 2025openai/codex

More with Claude Code

Reported by source

Review pull requests automatically with Claude in GitHub Actions

Install the Claude Code GitHub Action so Claude reviews each pull request with inline comments and responds to @claude requests.

code.claude.comanthropics/claude-code
Reported by source

Trace a disk-filling production bug from core dumps with an agent

B12 CTO Adam Marcus used Claude Code to analyze gigabyte core dumps filling web server disks and traced them to an RSS endpoint reading an endless radio stream.

blog.marcua.net · 28 Jan 2026anthropics/claude-code
Reported by source

Give Claude your Google Calendar and Gmail through a self-run MCP server

The open-source Google Workspace MCP server exposes Calendar, Gmail and other Workspace tools to Claude Desktop, Claude Code and other MCP clients, using your own OAuth client and optional read-only mode.

github.com

View all: Claude Code