SciClaw
01

Getting Started

1. What SciClaw Is

SciClaw is an AI co-worker built for scientific research. It helps researchers continuously carry out knowledge accumulation, task execution, output generation, and workflow improvement around a project, forming a truly closed-loop research workflow.
It is not just a chat assistant that answers questions, nor simply a bundle of isolated tools. It is a research collaboration system centered on projects, driven by tasks, and oriented toward outcomes.
Through its Proactive Agent Framework, SciClaw keeps literature search, data analysis, deep research, long-running computation, result organization, and experimental execution connected as an ongoing workflow. This allows research work to move from step-by-step manual coordination toward active, system-level collaboration.
For researchers, SciClaw's value is not just improving efficiency on individual tasks. More importantly, it helps teams accumulate research know-how, reuse proven methods, and reduce repetitive work, while steadily improving the quality and speed of project execution.

Scientific work today often runs into a few recurring problems:

  • Materials and results are scattered across files, chats, spreadsheets, and tools, making it difficult to build a unified project context
  • Many analysis, computation, search, and organization tasks recur repeatedly, but are hard to turn into reusable workflows
  • Preparing research outputs still depends on a large amount of manual copying, formatting, and rearranging, which is time-consuming and error-prone
  • Long-running and complex tasks require researchers to constantly monitor progress and manually connect the next step
  • Wet lab work, dry lab work, simulation, and data analysis often remain disconnected instead of forming a complete loop

SciClaw was designed to address exactly these problems by giving researchers a collaborative system that can understand projects, call the right capabilities, generate outputs, support reflection, and connect further into experimental execution.

2. Initial Setup

When you use SciClaw for the first time, the system starts with a short conversation to understand your research focus, working style, and usage preferences. Based on this, it initializes your AI Persona.

The purpose of this setup is to help SciClaw work in a way that better fits your habits during later conversations and task collaboration. After setup is complete, SciClaw generates an initial set of persona files based on your input to record how it should understand you, how it should position itself, and how it should collaborate with you.

Once your AI Persona is initialized, SciClaw can:

  • Understand your research background and task context more quickly
  • Communicate in a way that better fits your habits
  • Maintain a more consistent collaboration style across ongoing projects
  • Reduce the need to repeatedly explain your preferences and working style

This makes SciClaw more than a tool that simply answers questions. It becomes a collaborator that can continuously understand you and adapt to the way you work.

Your AI Persona is meant to evolve, not stay fixed forever.As you use SciClaw more, you can revisit and adjust it at any time so it better reflects your habits, research priorities, or team workflow. For details, see 06 AI Persona.

3. How SciClaw Works

A typical SciClaw workflow starts from a project, uses conversations and skills to launch work, turns complex requests into tasks, and then brings the results back into the project Library for later reuse.

  1. 1
    Create a project and upload relevant files or background materials into its Library
  2. 2
    Ask questions, analyze files, or trigger skills in conversation
  3. 3
    Let SciClaw answer directly in chat or create a task for more complex execution
  4. 4
    Review the task conclusion and generated files after execution finishes
  5. 5
    Use Foundry to turn accumulated project results into reports, presentations, posters, or structured data
SciClaw · USER GUIDE