EvoScientist/EvoSkills

🧬 Extend EvoScientist with Installable Skill & Knowledge Packs

What it solves

EvoSkills provides a comprehensive set of installable knowledge packs designed to extend AI agents (specifically EvoScientist) with domain-specific expertise for the entire academic research lifecycle. It transforms a general-purpose coding agent into a specialized scientific researcher capable of handling everything from initial ideation and literature review to experiment execution, paper writing, and peer-review rebuttals.

How it works

The project operates as a library of "skills" that can be installed into agents via commands or the skills.sh framework. These skills implement structured, multi-stage workflows for specific research tasks. For example, the research-ideation skill uses persona-driven refinement and Elo tournament ranking to select the best research ideas, while the experiment-pipeline enforces a 4-stage execution framework with attempt budgets to prevent "rabbit holes."

Central to the system is evo-memory, a persistent research memory layer that tracks successful and unsuccessful directions across research cycles, allowing the agent to evolve its strategies over time.

Who it’s for

  • AI Researchers: Those using AI agents to automate or augment the scientific discovery process.
  • Coding Agent Users: Users of Claude Code, Cursor, or Gemini CLI who want to add academic research capabilities to their tools.
  • Academic Writers: Researchers needing structured assistance with paper planning, drafting, and the rebuttal process.

Highlights

  • End-to-End Research Pipeline: Covers ideation, planning, experimentation, writing, and review.
  • Self-Evolution: Uses evo-memory to distill reusable patterns and avoid repeating past mistakes.
  • Rigorous Experimentation: Implements a 4-stage pipeline with strict attempt budgets and structured logging.
  • Academic Tooling: Includes specialized tools for publication-ready Matplotlib figures, Mermaid-based research lineage maps, and LaTeX paper drafting.
  • Mathematical Rigor: Features a Tao-style Olympiad proof workflow with calibrated abstention to avoid "hallucinated" proofs.

Related

  • Project
    Yuan1z0825/nature-skills

    A library of 22 packaged AI-agent workflows ("skills") for academic research — paper reading, Nature-style writing, figure generation, reviewer responses, literature search, and patent drafting — installable into Claude Code, Codex, Chatbox, and other agents.

  • Project
    LeonChaoX/qinyan-academic-skills

    A library of 187 installable AI agent skills for academic research, featuring a specialized suite for Nature-style scientific writing and manuscript preparation.

  • Project
    EvoScientist/EvoScientist

    EvoScientist is a Python framework that orchestrates a team of LLM‑driven agents to perform the full scientific workflow—planning, researching, coding, debugging, analysing data, and writing papers. It features self‑evolving memory, automatic skill extraction, multi‑provider model support, and can be accessed via a desktop Web UI, CLI/TUI, or chat platforms. Designed for a human‑on‑the‑loop paradigm, it lets users schedule recurring research runs, review generated skills, and integrate community skill packs, all under an Apache‑2.0 license.

  • Project
    K-Dense-AI/scientific-agent-skills

    A library of 177 scientific and research skills that provide curated documentation and workflows for AI agents to interact with scientific databases, Python packages, and research methodologies.

  • Project
    sentient-agi/EvoSkill

    A toolkit for automatically discovering and improving AI skills and prompts for coding agents, evolving general agents into specialists using benchmark-driven feedback.