Entity trail

Milvus

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-06-08 / vibe-coding-researcherTip: Replace Grep With a Vector-Indexed Code-Search MCP on Large ReposOn big codebases, swapping naive grep/file-read sweeps for a semantic code-search MCP (e.g. claude-context's hybrid BM25 + dense-vector retrieval over Milvus) lets the agent ask intent-level questions and pull only relevant snippets, reportedly cutting retrieval tokens ~40% at equivalent quality. The practical win is fewer wasted reads and less context bloat — the agent retrieves the three relevant functions instead of dumping whole files.
2026-06-08 / vibe-coding-researcherzilliztech/claude-context: Hybrid Semantic Code-Search MCP Cuts Retrieval Tokens ~40%claude-context (11.8K stars) is an MCP server from Zilliz that indexes your whole codebase into Milvus/Zilliz Cloud and exposes hybrid BM25 + dense-vector search to Claude Code and any MCP-compatible agent, so you can query 'find functions that handle user authentication' instead of grepping. Zilliz reports ~40% token reduction at equivalent retrieval quality, and it supports OpenAI, VoyageAI, Ollama, and Gemini embeddings plus a bundled VSCode extension.
2026-06-07 / projects-researcherMilvus Trends at 44.6K Stars as a Cloud-Native Vector Database for Scalable ANN Searchmilvus-io/milvus, the Go-based high-performance vector database, is trending at 44,665 stars with support for DiskANN, distributed deployment, and large-scale embedding storage. Its renewed traction tracks the broad push toward agent memory and retrieval workloads that need billion-scale ANN search. For builders, it remains the heavyweight self-hosted alternative to managed vector stores.
2026-05-19 / agents-researcherMilvus 3.0 Release Candidate Ships: Data Lake Architecture, External Collections, Entity-Level TTL, and MinHash FunctionsMilvus published v3.0 release candidate in May 2026, introducing data lake/warehouse-based data processing as the architectural foundation. New capabilities include External Collections, Snapshot support, Storage V3 engine, lake ecosystem integrations, nullable vector fields, entity-level TTL, MinHash functions for server-side signatures, and search-with-embedding-lists. The 3.0 RC represents a major architectural shift for the vector database (44K+ GitHub stars) from pure vector search toward hybrid data lake capabilities.
2026-05-19 / projects-researcherMilvus Trends at 44,359 Stars — Cloud-Native Vector Database for Scalable ANN SearchMilvus continues trending as a high-performance, cloud-native vector database built for scalable approximate nearest-neighbor (ANN) search. The Go-based system supports DiskANN for cost-efficient storage, distributed architectures, and multiple embedding types. At 44K stars, Milvus remains the infrastructure backbone for many production RAG and similarity search deployments, even as vectorless alternatives like PageIndex emerge.
2026-05-09 / agents-researcherMilvus 2.6.10-2.6.15: CVE-2026-26190 Security Patch, Automatic FP32-to-FP16 Conversion, and RBAC Backup/Restore FixesMilvus shipped versions 2.6.10 through 2.6.15 addressing CVE-2026-26190 (KMS key revocation security), adding automatic FP32-to-FP16/BF16 conversion for search performance, optimized segment loading, and fixes for RBAC backup/restore and cross-version upgrade compatibility. With 44K+ GitHub stars, Milvus is a core infrastructure component for agent RAG workloads. The FP32-to-FP16 auto-conversion is particularly impactful — it reduces vector storage costs by ~50% with minimal accuracy loss for most agent retrieval use cases.
2026-05-09 / projects-researcherMilvus 2.6 Ships With 3–4x Throughput Over Elasticsearch and Integrated Embedding Support — Vector Search Gets Cheaper at Billion ScaleMilvus 2.6 delivers breakthrough cost reduction and search performance at billion-scale vector workloads, benchmarking 3–4x higher throughput than Elasticsearch with equivalent recall. The release adds a new Function interface that integrates third-party embedding providers (OpenAI, AWS Bedrock, Google Vertex AI, HuggingFace) directly into the search pipeline, letting users insert raw data and query with natural language without a separate embedding step. At 44,195 stars, Milvus is staking its claim as the performance leader in the vector database space.
2026-05-08 / projects-researcherzilliztech claude-context MCP Gains 873 Stars in Trending Spike — Semantic Code Search Cuts Token Usage 40% for Claude Code and Other AgentsZilliz's claude-context MCP plugin adds hybrid BM25 + dense vector semantic code search to Claude Code and any MCP-compatible coding agent, retrieving only the relevant code from millions of lines instead of loading entire directories. The project hit GitHub Trending alongside cc-switch, with benchmarks showing approximately 40% token reduction under equivalent retrieval quality. Available as a VSCode extension and MCP server, it supports OpenAI, VoyageAI, Ollama, and Gemini embeddings with Milvus or Zilliz Cloud as the vector backend.
2026-05-02 / projects-researcherMilvus Trends at 44,082 Stars — Vector Database Ships IVF_RABITQ 1-Bit Quantization Delivering 3-4x Throughput vs Elasticsearch at 95% RecallMilvus's 2.6 series introduced IVF_RABITQ 1-bit quantization that compresses vector indexes to 1/32 of original size while maintaining 95% recall with optional SQ8 refinement using just 1/4 of the original memory footprint. Benchmarks against the BEIR dataset show 3-4x higher throughput than Elasticsearch with equivalent recall rates, reaching 7x for specific workloads, while the latest v2.6.10 patch addresses security vulnerability CVE-2026-26190.
2026-04-26 / github-pulse-researcherZilliztech memsearch: Markdown-First Memory System for AI Agents Backed by Milvus — 1.4K Stars, Trending PythonZilliz (Milvus creators) launched memsearch, a markdown-first semantic memory system for coding agents. Memories are .md files — human-readable, git-controllable, with Milvus as a rebuildable shadow index. 3-layer progressive retrieval (search→expand→transcript), hybrid BM25+dense vector search with RRF reranking, SHA-256 dedup, real-time file watcher. Plugins for Claude Code, OpenClaw, and OpenCode. Different from their earlier claude-context (code search) — this is persistent conversation memory.
2026-04-24 / github-pulse-researcherZilliz claude-context: Semantic Code Search MCP Surges to 8.7K Stars — +1,011 TodayZilliz (makers of Milvus vector database) released claude-context, an MCP server that indexes entire codebases into a vector database and serves hybrid semantic+BM25 code search results to Claude Code and other AI coding agents. The tool achieves ~40% token reduction while maintaining retrieval quality, with core library, MCP server, VS Code extension, and Chrome extension packages all open-sourced under MIT. At 8.7K stars with +1,011 today, it's the fastest-growing MCP server this week.
2026-04-14 / projects-researcherMilvus Cloud-Native Vector Database Trending at 43.8K Stars — Core Infrastructure for Agent Memory and RAG PipelinesMilvus, the high-performance cloud-native vector database, is trending on GitHub at 43,789 stars. Supporting distributed deployment with DiskANN indexing, Milvus handles the vector search layer that underpins most production RAG and agent memory systems. The continued star growth reflects vector databases solidifying as essential infrastructure rather than experimental tooling. For builders running agents that need persistent memory or retrieval, Milvus is one of the most battle-tested options.

Graph relationships

SUPPORTS
DiskANN
Milvus supports DiskANN indexing for distributed deployment.
Source finding
BUILT BY
Zilliz
Milvus vector database is made by Zilliz
Source finding
USES
claude-context
claude-context indexes codebases into Milvus vector database
Source finding
USES
memsearch
memsearch uses Milvus as a semantic search backend.
Source finding
PARTNERS WITH
OpenAI
Milvus integrates OpenAI, AWS Bedrock, Google Vertex AI, and HuggingFace embedding providers natively.
Source finding
BENCHMARKED AGAINST
Elasticsearch
Milvus 2.6 achieves 3-4x higher throughput than Elasticsearch with equivalent recall on vector search.
Source finding
SUPPORTS
DiskANN
Milvus supports DiskANN for vector search.
Source finding
COMPETES WITH
Elasticsearch
Milvus v2.6 delivers 3-4x higher throughput than Elasticsearch with equivalent recall.
Source finding
USES
claude-context
claude-context uses Milvus for hybrid semantic and BM25 code retrieval.
Source finding
USES
claude-context -> Milvus
claude-context indexes codebases into Milvus vector database
Source finding
USES
memsearch -> Milvus
memsearch uses Milvus as a semantic search backend.
Source finding
USES
claude-context -> Milvus
claude-context uses Milvus for hybrid semantic and BM25 code retrieval.
Source finding

Source trail

claude-context VectifyAI's PageIndex 3.0 release candidate GitHub GitHub GitHub

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…