MCP server for prismAId AI-assisted systematic reviews and protocol conformance checking.
This MCP server provides tools for prismAId, an AI-assisted workflow for systematic literature reviews. It uses generative AI models to streamline screening and analysis, and supports protocol conformance checking alongside review activities. The project is positioned for open science and open-source collaboration.
🛠️ Key Features
MCP server for prismAId AI-assisted systematic reviews
Generative AI–based screening and analysis of research papers
Protocol conformance checking
🚀 Use Cases
Streamlining systematic literature reviews
Supporting screening of scientific papers
Verifying conformance to review protocols during workflow execution
⚡ Developer Benefits
Replicable review methods
Simple-to-use tools that do not require coding skills
⚠️ Limitations
The available description does not specify tool names, configuration options, or the full set of MCP tools.
prismAId offers a comprehensive set of tools for systematic literature reviews:
Core Tools
Screening - Filter and tag manuscripts to identify items for exclusion
Download - Download papers from Zotero collections or from URL lists
Convert - Convert files (PDF, DOCX, HTML) to plain text for analysis
Review - Process systematic literature reviews based on TOML configurations
RevAIse documentation support - Optionally document review stages as RevAIse review records
Workflow
Our tools support a comprehensive systematic review workflow following the standard sequence: Search → Screen → Download → Convert → Review. RevAIse support can document Zotero download, screening, and review/extraction stages in one cumulative review record.
Access Methods
AI agents via the MCP server - A main entry point: connect an AI assistant to the prismAId MCP server and drive every tool in conversation
Command Line Interface - For users who prefer terminal-based workflows
Web Initializer - A browser-based setup tool for configuring reviews
Programming Libraries - API access through multiple languages:
Go (native implementation)
Python package
R package
Julia package
Specifications
Review protocol: Supports any literature review protocol with a preference for PRISMA 2020, which inspired our project name.
Review documentation: Optional RevAIse review-record support with cumulative updates and automatic backups; see the RevAIse integration guide.
Protocol conformance: Check RevAIse review records against reporting protocols such as PRISMA 2020, and get a protocol's full requirement checklist, using the SHACL shapes published by RevAIse; see the conformance and guidance docs.
Cohere: Command, Command Light, Command R, Command R+, Command R7B, Command R (August 2024), Command A, and Command A Reasoning
Anthropic: Claude 3 Sonnet, Claude 3 Opus, Claude 3 Haiku, Claude 3.5 Haiku, Claude 3.5 Sonnet, Claude 3.7 Sonnet, Claude 4.0 Sonnet, Claude 4.0 Opus, Claude 4.5 Opus, Claude 4.5 Sonnet, and Claude 4.5 Haiku
DeepSeek: DeepSeek Chat v3, and DeepSeek Reasoner v3
Perplexity: Sonar, Sonar Pro, Sonar Reasoning Pro, and Sonar Deep Research
Cloud Providers: AWS Bedrock, Azure AI, Vertex AI
Self-Hosted: OpenAI-compatible endpoints
Screening capabilities: Deduplication, language filtering, article type classification, and off-topic detection
Output format: Data in CSV or JSON formats
Performance: Efficiently processes extensive datasets with minimal setup and no coding required
Programming Language: Core implementation in Go with bindings for Python, R, and Julia
This project was initiated with the generous support of a SIS internal project from NILU. Their support was crucial in starting this research and development effort. Further, acknowledgment is due for the research credits received from the OpenAI Researcher Access Program and the Cohere For AI Research Grant Program, both of which have significantly contributed to the advancement of this work.
prismAId
