Observed Signal · Aug 18, 2026 · Technical Release · Source: DEV Community · Impact: 4/5 · Sentiment: Positive
Connect Foundry IQ Knowledge Base to LangGraph via MCP
Technical how-to explaining how to ground a LangGraph agent in Microsoft Foundry IQ agentic retrieval by calling the knowledge base's MCP endpoint. The guide covers architecture, API-version differences (2026-04-01 vs 2026-05-01-preview), authentication and token refresh patterns, preserving citations in LangGraph state, per-user permission headers, and design guidance on planner handoffs. It includes code samples (Python) for creating knowledge sources/bases, verifying retrieval via the SDK, implementing an httpx.Auth that refreshes Azure bearer tokens, loading the MCP tool into LangGraph, parsing MCP tool results, and enforcing user-scoped permissions via the x-ms-query-source-authorization header.
Provides operationally critical guidance for integrating enterprise agent frameworks with Microsoft Foundry IQ via MCP, covering API-version tradeoffs, token refresh, per-user permission enforcement, and citation preservation — all important for production-grade agent deployments.
Track Microsoft Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Microsoft Foundry IQ exposes each knowledge base as an MCP server offering a single tool named knowledge_base_retrieve.
- Agentic retrieval REST API GA is 2026-04-01; 2026-05-01-preview adds answer synthesis, configurable reasoning effort, document-level permissions and sensitivity-label metadata.
- MCP tool results return grounding data as a JSON-encoded string inside a text block (result.content[0].text), and do not include the SDK 'activity' or 'references' arrays.
- Bearer tokens expire; the tutorial demonstrates using a custom httpx.Auth (EntraBearerAuth) that calls azure-identity's token provider per request to avoid frozen tokens.
- Per-user permission filtering requires passing the end user's token in the x-ms-query-source-authorization header at query time (distinct from the service credential).
Connected Companies & Entities
5 Entities mapped“A Microsoft Foundry project and resource, with an LLM deployment (e.g. `gpt-5-mini`) and an embedding model (e.g. `text-embedding-3-large`)....”
“A Microsoft Foundry project and resource, with an LLM deployment (e.g. `gpt-5-mini`) and an embedding model (e.g. `text-embedding-3-large`)....”
“Any MCP-compatible client can call it — including LangGraph, via `langchain-mcp-adapters`....”
“The same knowledge base you just built is reachable from Microsoft Agent Framework, Foundry Agent Service, GitHub Copilot, Claude and Cursor...”
“The same knowledge base you just built is reachable from Microsoft Agent Framework, Foundry Agent Service, GitHub Copilot, Claude and Cursor...”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
LangGraph vs Microsoft Agent Framework: State‑First or Discover‑Later
This technical comparison contrasts LangGraph and Microsoft Agent Framework (MAF) by how each handles workflow state in agentic systems. LangGraph enforces a schema-first design: developers declare a typed state contract up front, compile a StateGraph, and use an interrupt/checkpointer model to serialize full state and resume execution exactly where it paused. MAF separates Agent behaviour from Workflow, relies on message passing (no global state schema), and handles human-in-the-loop pauses via an emit-and-rerun request/response model. The article argues LangGraph imposes upfront cost but yields clearer, more maintainable long-running workflows, while MAF enables faster prototyping but can complicate composition and consistent state management in complex production systems. It also notes ecosystem differences: LangGraph is Python-first with a larger open-source community; MAF is Microsoft-backed with Azure integrations and migration paths from AutoGen and Semantic Kernel. (Published 2026-04-28.)
Building Semantic Search with GPT-5 and Microsoft Foundry
This technical tutorial demonstrates building a production-grade semantic search pipeline using GPT-5 for query planning and Microsoft Foundry's Foundry IQ (an agentic retrieval layer on Azure AI Search) for retrieval. The guide walks through creating a knowledge source (backed by blob storage and automatically indexed by Azure AI Search), configuring a knowledge base that uses an LLM deployment (example: gpt-5-mini) for query planning and answer synthesis, wiring the knowledge base into a Foundry agent via an MCP endpoint, and tuning retrieval_reasoning_effort to balance cost, latency, and answer quality. The article emphasizes multi-hop queries, iterative planning, and the trade-offs between extractive and synthesized answers.
Model Context Protocol (MCP) Fundamentals Guide
This technical tutorial introduces the Model Context Protocol (MCP), an open standard for connecting large language models (LLMs) to external tools, data sources and services. It demonstrates building an 'Analyzer' MCP server using the FastMCP framework, explains MCP message types (tool discovery and tool execution), and shows transports (stdio, HTTP, WebSockets). The post describes moving from local development to production via Bedrock AgentCore Runtime—containerizing MCP servers, registering them with a runtime client, and securing access with Amazon Cognito. It also shows how Strands Agents can consume remote MCP tools as if local, and outlines best practices: descriptive docstrings, strict Python type hints, error handling, logging, and composable tool design to enable chaining and context awareness. The article targets developers building reusable, secure, scalable agent-accessible tools across multiple LLMs (e.g., Claude, GPT, Nova).
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.