Download as PDF, PPTX
![Resources
Resources are identified using URIs that follow this format:
[protocol]://[host]/[path]
For example: alfresco:///home/user/documents/report.pdf
• Text resources contain UTF-8 encoded text data. These are suitable
for source code, configuration files, log files, JSON/XML data
• Binary resources contain raw binary data encoded in base64. These
are suitable for Images, PDFs, Audio files, Video files
MCP](https://image.slidesharecdn.com/alfresco-ttl164-mcp-250127135734-705dfd0f/85/Alfresco-and-the-Model-Context-Protocol-MCP-19-320.jpg)
![Resources
Resources are identified using URIs that follow this format:
[protocol]://[host]/[path]
For example: alfresco:///home/user/documents/report.pdf
• Text resources contain UTF-8 encoded text data. These are suitable
for source code, configuration files, log files, JSON/XML data
• Binary resources contain raw binary data encoded in base64. These
are suitable for Images, PDFs, Audio files, Video files
MCP](https://image.slidesharecdn.com/alfresco-ttl164-mcp-250127135734-705dfd0f/75/Alfresco-and-the-Model-Context-Protocol-MCP-19-2048.jpg)
Uploaded byAngel Borroy López
Alfresco and the Model Context Protocol (MCP)
The document is a comprehensive guide on integrating Alfresco with external REST APIs, detailing configuration of custom web scripts, authentication, and document synchronization. It introduces the Model Context Protocol (MCP) for standardized communication with AI models, providing architecture details and examples of capabilities like resources, prompts, and tools. The document also discusses semantic search technologies, vector databases, and various large language models (LLMs) in the context of retrieval-augmented generation (RAG).
More Related Content
Similar to Alfresco and the Model Context Protocol (MCP)
Alfresco and the Model Context Protocol (MCP)
- 2. January 23, 2025 ©2024Hyland Software, Inc. and its affiliates. All rights reserved. All Hyland product names are registered or unregistered trademarks of Hyland Software, Inc. or its affiliates in the United States and other countries. Alfresco and the Model Context Protocol (MCP) Alfresco TTL#164 Angel Borroy Developer Evangelist
- 3. • Semantic Search •Retrieval-Augmented Generation (RAG) • Model Context Protocol (MCP) • Alfresco & MCP Agenda
- 4.
- 5. Semantic Search How canI integrate Alfresco with an external REST API? 1 2 3 4 5 Guide to developing custom integrations with Alfresco Using Web Scripts to extend Alfresco functionality Alfresco API Authentication methods explained Alfresco Integration: External application use cases Configuring Alfresco to Consume External APIs Synonyms Conceptual Relationships Contextual Relevance Broader Understanding Connecting Alfresco to external systems and consuming external services Authentication, Web Scripts and Use Cases Development and configuration Developer and Admin focused solutions
- 6. Semantic Search Knowledge Base Alfresco Embedding 0.12,-0.34, 0.56, 0.78, -0.90, 0.23 Storage Vector Database Similarity Search Vector Database Documents Content associated with search
- 7. Semantic Search Embedding Library SentenceTransformersTransformers spaCy Nomic Type Sentence embeddings NLP framework Lightweight NLP Embedding Embedding Type Dense vectors Dense vectors, word, or token Word, document Dense vectors Ease of Use High Medium to High Very High High Dimensions 384 to 1024 768 to 1024 300 (default) 768 (default) Use Cases Similarity, search Broad NLP tasks Lightweight NLP Quick embedding Cons Resource-intensive High memory usage Static embeddings API Dependency License Apache License 2.0 Apache License 2.0 MIT MIT
- 8. Semantic Search Vector Database WeaviateMilvus Vespa Qdrant OpenSearch Elasticsearch Type Database Database Engine Database Search Engine Search Engine Indexing HNSW IVF, HNSW k-NN HNSW k-NN k-NN Ease of Use Easy Moderate Complex Easy Moderate Easy Use Case RAG Semantic search RAG Recommendations Real-time Recommendations Search Recommendations Hybrid search Analytics Hybrid search Analytics Cons Schema management Complex setup Steeper learning curve Still maturing Slower on high- dimension vectors Less optimized for vector-only queries License BSL 1.1 Apache License 2.0 Apache License 2.0 Apache License 2.0 Apache License 2.0 Elastic License 2.0
- 9.
- 10.
- 11. RAG How can Iintegrate Alfresco with an external REST API? 1 2 3 4 5 Guide to developing custom integrations with Alfresco Using Web Scripts to extend Alfresco functionality Alfresco API Authentication methods explained Alfresco Integration: External application use cases Configuring Alfresco to Consume External APIs LLM Response To configure Alfresco for integration with an external REST API, you can either expose custom Web Scripts or consume external APIs directly. For exposing functionality, create and deploy Web Scripts using Java or JavaScript, enabling RESTful endpoints to interact with your external systems. If consuming APIs, configure HTTP connectors in Alfresco to make secure calls to external APIs. Authentication is key—ensure you have configured OAuth2 or Basic Authentication for secure access to Alfresco's APIs. Additionally, check that your endpoints are properly formatted, and use tools like Postman to test API requests. For document synchronization, use Alfresco’s REST API to retrieve documents, metadata, or folder structures.
- 12. RAG User Query What isContent Federation? Query Embedding 0.12, -0.34, 0.56, 0.78, -0.90, 0.23 Similarity Search Vector Database Contextual Data Content associated with retrieval Response Generation LLM
- 13. RAG LLM (Large LanguageModel) Mistral DeepSeek Llama 3 Gemma 2 Qwen-2.5 Phi-3.5 Memory Efficiency High Moderate High High High Very High Inference Speed Fast with GPU Moderate Fast with GPU Moderate Very Fast Fast with GPU Parameter Size 7B 14B 13B 10B 8B 7B Features Efficient for text generation and summarization, retains context well Reasoning model with high context retention for RAG More accurate and nuanced; useful for detailed responses in RAG tasks Optimized for memory usage and contextual accuracy in RAG applications Strong in multilingual contexts, optimized for efficient deployment Focuses on low latency, lightweight and memory- efficient; suitable for smaller RAG tasks License Apache License 2.0 DeepSeek License Agreement LLaMA Community License Apache License 2.0 Apache License 2.0 MIT
- 14.
- 15.
- 16. Background & Motivation Anopen standard designed to unify how applications communicate context to AI models MCP MCP Server https://github.com/modelcontextprotocol/servers
- 17. Architecture • MCP Hosts:Programs like Claude Desktop, IDEs, or AI tools that want to access data through MCP • MCP Clients: Protocol clients that maintain 1:1 connections with servers • MCP Servers: Lightweight programs that each expose specific capabilities through the standardized Model Context Protocol • Local Data Sources: Your computer’s files, databases, and services that MCP servers can securely access • Remote Services: External systems available over the internet (e.g., through APIs) that MCP servers can connect to MCP
- 18. MCP Server Choose yourflavor: • Python SDK: https://github.com/modelcontextprotocol/python-sdk • TypeScript SDK: https://github.com/modelcontextprotocol/typescript-sdk MCP servers can provide three main types of capabilities: • Resources • File-like data that can be read by clients (like API responses or file contents) • Prompts • Pre-written templates that help users accomplish specific tasks • Tools • Functions that can be called by the LLM (with user approval) MCP
- 19. Resources Resources are identifiedusing URIs that follow this format: [protocol]://[host]/[path] For example: alfresco:///home/user/documents/report.pdf • Text resources contain UTF-8 encoded text data. These are suitable for source code, configuration files, log files, JSON/XML data • Binary resources contain raw binary data encoded in base64. These are suitable for Images, PDFs, Audio files, Video files MCP
- 20. Prompts Prompts in MCPare predefined templates that can: • Accept dynamic arguments • Include context from resources • Chain multiple interactions • Guide specific workflows • Surface as UI elements (like slash commands) MCP
- 21. Tools Tools in MCPallow servers to expose executable functions that can be invoked by clients and used by LLMs to perform actions. • Discovery: Clients can list available tools through the tools/list endpoint • Invocation: Tools are called using the tools/call endpoint, where servers perform the requested operation and return results • Flexibility: Tools can range from simple calculations to complex API interaction MCP
- 22. Transport MCP uses JSON-RPC2.0 as its wire format (request, response and notification) Built-in Transport Types: • Standard Input/Output (stdio) for command-line tools • Server-Sent Events (SSE) for server-to-client streaming MCP allows custom transport implementations MCP
- 23. MCP Client • Connectto the specified server • List available tools • Start an interactive chat session where you can: • Enter queries • See tool executions • Get responses from LLM Some frameworks and apps implementing MCP Client: • langchain4j, spring-ai-mcp, Claude Desktop App MCP
- 24. Claude APP • Setup Git MCP Agentic Tooling with Claude Desktop MCP
- 25.
- 26. Alfresco & MCP Alfresco MCPServer MCP Client LLM TypeScript SDK https://github.com/aborroy/alfresco-mcp-poc