SCP: Accelerating Discovery with a Global Web of Autonomous Scientific Agents

TL;DR

The paper presents the Science Context Protocol (SCP), an open-source framework that accelerates discovery by unifying scientific resources (tools, models, data, instruments) and by orchestrating end-to-end experiment lifecycles through a centralized SCP Hub and federated edge Servers. It introduces a rich, JSON-based protocol for detailed experiment planning, context, provenance, and governance, enabling secure, multi-institution collaboration between AI agents and human researchers. Core contributions include the hub-and-spoke architecture, intelligent task planning, fine-grained workflow specifications, robust execution with validation and rollback, and native wet-lab integration, demonstrated via case studies spanning automated protocol design, PDF-to-protocol reproduction, AI-driven drug discovery, and dry–wet protein engineering. The platform’s open-source Intern-Discovery tool ecosystem, with over 1,600 tools, underpins scalable, reproducible, and auditable multi-agent science, with potential to substantially reduce integration overhead and enhance reproducibility across disciplines and institutions.

Abstract

We introduce SCP: the Science Context Protocol, an open-source standard designed to accelerate discovery by enabling a global network of autonomous scientific agents. SCP is built on two foundational pillars: (1) Unified Resource Integration: At its core, SCP provides a universal specification for describing and invoking scientific resources, spanning software tools, models, datasets, and physical instruments. This protocol-level standardization enables AI agents and applications to discover, call, and compose capabilities seamlessly across disparate platforms and institutional boundaries. (2) Orchestrated Experiment Lifecycle Management: SCP complements the protocol with a secure service architecture, which comprises a centralized SCP Hub and federated SCP Servers. This architecture manages the complete experiment lifecycle (registration, planning, execution, monitoring, and archival), enforces fine-grained authentication and authorization, and orchestrates traceable, end-to-end workflows that bridge computational and physical laboratories. Based on SCP, we have constructed a scientific discovery platform that offers researchers and agents a large-scale ecosystem of more than 1,600 tool resources. Across diverse use cases, SCP facilitates secure, large-scale collaboration between heterogeneous AI systems and human researchers while significantly reducing integration overhead and enhancing reproducibility. By standardizing scientific context and tool orchestration at the protocol level, SCP establishes essential infrastructure for scalable, multi-institution, agent-driven science.

Figures (7)

• Figure 1: SCP overview. The Science Context Protocol (SCP) is an open-source standard specifically designed to accelerate scientific discovery. By establishing a standardized connectivity framework, it enables efficient interaction between discovery-oriented applications and external research assets—such as laboratory instruments, databases, knowledge repositories, large language models (LLMs), specialized computational models, tools, and APIs. SCP aims to foster a hybrid dry-wet, multi-institution collaborative research paradigm and serve as a novel support platform to enable the collaborative evolution of researchers, research tools, and research subjects in a new era of multi-agent-driven scientific investigation and discovery.
• Figure 2: SCP architecture overview. The SCP Hub coordinates interactions between user-facing clients (top) and various SCP edge servers (bottom) that interface with laboratory instruments, databases, AI models, and other tools. Researchers interact with the system through an SCP client application, which communicates with the Hub. The Hub manages experiment context, planning, and task scheduling across the network of tools. Each SCP edge server registers available devices or services with the Hub, executes tasks on those resources, and streams results back to the Hub in real time. This design enables a seamless flow of information and commands between human researchers, AI-driven agents, and physical lab equipment under a unified protocol.
• Figure 3: Distribution of disciplines and functions of tools available on the Intern-Discovery platform intern-discover-platform. The tool collection is continuously updated and expanded.
• Figure 4: Case Study 1: Automated Experimental Protocol Design and Execution.
• Figure 5: Case Study 2: Automated Reproduction of an Existing Protocol from PDF.