
from neurico120
Orchestrates end-to-end AI research: literature review, experiment execution, analysis, and paper drafting from a structured idea spec.
NeuriCo automates the scientific research lifecycle: given a structured idea (YAML with title, domain, hypothesis), it runs literature review, designs experiments, executes code, collects results and plots, and drafts a LaTeX paper and GitHub repo. It aims to produce reproducible experiment artifacts and a paper draft from a research idea.
Use NeuriCo to prototype AI/ML research experiments, reproduce papers, or rapidly iterate on hypotheses where automation can speed up literature search, experiment orchestration, and paper drafting. Best when you have access to AI coding CLIs and sufficient compute (Docker recommended).
./neurico fetch|submit|run.Designed for agent-enabled coding CLIs like Claude Code, Codex, and Gemini CLI that can orchestrate multi-step experiments, run code inside Docker, and push results to GitHub.
NeuriCo is an autonomous AI research framework that takes a YAML research idea and orchestrates the full lifecycle: literature review, experiment design, code execution, analysis, and LaTeX paper drafting. The SKILL.md is well-structured with clear requirements, installation instructions (Docker and native), and input/output specs. No bundled scripts were present for runtime testing. Security section in the skill body mentions no secrets are uploaded and experiments run in Docker isolation, but the native install path includes `curl | sh` for uv installation which is a security concern.
Well-documented skill from ChicagoHAI (University of Chicago). The curl|sh pattern for uv installation is a standard practice but deducts from security score. The skill explicitly documents its security model (Docker isolation, secret filtering). Useful for AI researchers but niche audience. No scripts to execute or audit dynamically — scoring based on SKILL.md static analysis only.