forger-labs-hq

Researchforge — Development skill for Claude Code

Development community

A lab protocol for coding agents.

How to install Researchforge

This entry records only its repository, not the path inside it, so there is no exact command to give. Open forger-labs-hq/researchforge and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Researchforge does

A lab protocol for coding agents. Freeze baselines, run hypotheses in isolated worktrees, reject failures and validate improvements from Claude Code, Cursor or a direct API key (Gemini/Anthropic/openAI).

Alternatives in Development

Octo Freeze — Restrict file edits to a specific directory boundary 2.8k ★
Investigate Performance — Investigate a Nimbalyst performance problem (freeze, lag, idle CPU, slow op, memory growth) with measured evid 1.6k ★
Lab Skills — Skills distilled from the Lab's real work and collaboration practices 309 ★

README

ResearchForge

Your AI generates ideas. ResearchForge tests which ones actually hold up.

Claude Code or Cursor is the researcher. ResearchForge is the lab protocol.

From papers to proof.

▶ Watch the intro

Product introduction

ResearchForge turns a research question — or an "improve my repository" goal — into reproducible, traceable evidence. The agent proposes; ResearchForge freezes, executes, measures, rejects, and validates. It finds relevant papers, generates testable hypotheses, benchmarks competing implementations against a frozen baseline in isolated local workspaces, and delivers the strongest supported result as a clean branch, an engineering report, or a research package.

Works with **Claude Code** (slash-command skills), **Cursor** (MDC rules), or **standalone** with any AI API key — install once for your whole machine.

Inspired by Andrej Karpathy's [autoresearch](https://github.com/karpathy/autoresearch), where an agent autonomously runs training experime