What Microsoft's Foundry IQ Knowledge Graph update and real-world MCP OAuth headaches tell us about building self-hosted RAG The PipesHub team, building an open-source Apache 2.0 self-hosted RAG and context layer, reported that combining deterministic structural edges with selective entity linking avoids the noisy graphs and indexing costs of pure LLM triplet extraction. The team also implemented explicit service-account token marking (PR #6048) so RBAC and org-level usage tracking work for autonomous agents, and found that Dropbox's MCP OAuth discovery endpoint advertises a registration URL that returns 403 Forbidden, arguing pre-registered OAuth clients must be supported alongside DCR/CIMD. PipesHub said its stack, combining Qdrant with Neo4j/ArangoDB and 50+ connectors, ranked first on EnterpriseRAG-Bench at 93.24. Two things stood out this week if you build production RAG and agent infrastructure: 1. Microsoft added Knowledge Graphs and Microsoft-Retriever-1 to Foundry IQ InfoWorld, Nov 19 , specifically to stop developers from forcing large frontier models to plan multi-hop retrieval loops over flat vector chunks. 2. Starburst launched Context Studio Nov 18 , exposing governed semantic context over MCP so agents don't hallucinate conflicting business definitions. Both validate the same shift: vector search alone isn't enough, and stuffing everything into a frontier LLM prompt is too slow and expensive. At the same time, building an open-source, self-hosted alternative at PipesHub https://github.com/pipeshub-ai/pipeshub-ai https://github.com/pipeshub-ai/pipeshub-ai has taught us a few practical lessons that vendor announcements usually gloss over: - Knowledge graphs shouldn't rely purely on expensive LLM triplet extraction : Automated entity extraction over messy internal jargon creates noisy edges and huge indexing bills. Combining deterministic structural edges document - section/block - linked Jira ticket / GitHub PR / Slack thread with selective entity linking gives you multi-hop grounding and traceable citations without unpredictable cloud indexing fees. - Agent identity = human user identity : When an autonomous agent queries your context layer in the background, treating its token like a generic admin key breaks both auditability and billing. This week we implemented explicit service-account token marking PR 6048 so RBAC and org-level credit/usage tracking work cleanly for non-human actors. - MCP OAuth Dynamic Client Registration DCR is still rocky in the wild : While expanding our MCP server catalog adding Miro today and testing Dropbox's official MCP server , we hit a classic real-world spec gap: Dropbox's .well-known discovery endpoint advertises a registration URL, yet POST /oauth2/register returns 403 Forbidden dcr failed . Supporting pre-registered OAuth clients alongside CIMD/DCR is mandatory if you want MCP integrations to actually work in production. If you want a self-hosted, Apache 2.0 context layer Qdrant + Neo4j/ArangoDB + 50+ connectors + MCP + Python/TS/Go SDKs that ranked 1 on EnterpriseRAG-Bench 93.24 without Azure or lakehouse lock-in, check out the repo: For those running self-hosted MCP servers in production—how are you handling OAuth registration and service-account permissions?