Milvus
Cloud-native open-source vector database built for billion-scale similarity search and AI applications.
Milvus
Milvus is a purpose-built open-source vector database designed for high-performance similarity search at scale. It supports multiple index types (HNSW, IVF, FLAT, SCANN) and hybrid scalar+vector filtering, making it the vector layer for production RAG, recommendation, and multimodal search applications.
Key Features
- Billion-scale search: Horizontally scalable with independent compute and storage nodes
- Multiple index types: HNSW, DiskANN, IVF_FLAT, IVF_SQ8, SCANN for different speed/accuracy trade-offs
- Hybrid search: Combine dense vector ANN with sparse BM25 keyword and scalar field filters
- Multi-tenancy: Collection-level isolation with role-based access control
- Streaming ingestion: Real-time data ingestion via Kafka/Pulsar integration
- MilvusLite: Embedded mode that runs in-process for development and edge deployments
Quick Start
# Start Milvus standalone
wget https://raw.githubusercontent.com/milvus-io/milvus/master/scripts/standalone_embed.sh
bash standalone_embed.sh start
pip install pymilvus
from pymilvus import MilvusClient
client = MilvusClient("milvus_demo.db") # MilvusLite
client.create_collection("my_embeddings", dimension=768)
client.insert("my_embeddings", [{"id": 1, "vector": [0.1]*768}])
results = client.search("my_embeddings", data=[[0.1]*768], limit=5)
Add to ai-supply
npx ai-supply add milvus-vector-database
Curated mirror of the open-source Milvus (Apache-2.0). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/milvus-vector-database/check). Click a policy:
Consume Milvus programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/milvus-vector-database
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/milvus-vector-database/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add milvus-vector-database
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/milvus-vector-database/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "milvus-vector-database" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.