{"_id":"@agentxin-ai/plugin-milvus","name":"@agentxin-ai/plugin-milvus","dist-tags":{"latest":"0.0.3"},"versions":{"0.0.3":{"name":"@agentxin-ai/plugin-milvus","version":"0.0.3","type":"module","main":"./dist/index.js","module":"./dist/index.js","types":"./dist/index.d.ts","exports":{"./package.json":"./package.json",".":{"@agentxin-plugins-starter/source":"./src/index.ts","types":"./dist/index.d.ts","import":"./dist/index.js","default":"./dist/index.js"}},"dependencies":{"@zilliz/milvus2-sdk-node":"2.6.0","tslib":"^2.3.0"},"peerDependencies":{"@langchain/core":"^0.3.72","@nestjs/config":"^4.0.2","@nestjs/common":"^11.1.6","@agentxin-ai/plugin-sdk":"^3.6.1","zod":"^3.25.67","chalk":"^4.1.2","uuid":"^8.3.2"},"_id":"@agentxin-ai/plugin-milvus@0.0.3","description":"This package exposes the Milvus vector store integration for the [AgentXinAI](https://github.com/agentxin-ai/agentxin) platform. It wraps the official `@zilliz/milvus2-sdk-node` client with an opinionated LangChain-compatible adapter that promotes frequen","_nodeVersion":"22.22.0","_npmVersion":"10.9.4","dist":{"integrity":"sha512-fjDbftgrIdIGdOafXAG/rLfiDeR+QFlfN5YefOb4ycqMdASO2nJ5Fwj7RgUc5qaN2ahBkmGe8VA5IRRR0H88Dg==","shasum":"7122e525122a05add9381a8a77f32a29decf4897","tarball":"https://registry.npmjs.org/@agentxin-ai/plugin-milvus/-/plugin-milvus-0.0.3.tgz","fileCount":17,"unpackedSize":67718,"signatures":[{"keyid":"SHA256:DhQ8wR5APBvFHLF/+Tc+AYvPOdTpcIDqOhxsBHRwC7U","sig":"MEQCIFY2/gpSr1bIhaFKka5e91Xe6rK+ChLZGrsy42FAijI8AiACtSrPIjPlSWKXVdKF9wvYeEzLvrEqFUUoGJGbwCZkyA=="}]},"_npmUser":{"name":"agentxin-ai","email":"1304040880@qq.com"},"directories":{},"maintainers":[{"name":"agentxin-ai","email":"1304040880@qq.com"}],"_npmOperationalInternal":{"host":"s3://npm-registry-packages-npm-production","tmp":"tmp/plugin-milvus_0.0.3_1775181626782_0.39847701426844684"},"_hasShrinkwrap":false}},"time":{"created":"2026-04-03T02:00:26.667Z","0.0.3":"2026-04-03T02:00:26.940Z","modified":"2026-04-03T02:00:27.174Z"},"maintainers":[{"name":"agentxin-ai","email":"1304040880@qq.com"}],"description":"This package exposes the Milvus vector store integration for the [AgentXinAI](https://github.com/agentxin-ai/agentxin) platform. It wraps the official `@zilliz/milvus2-sdk-node` client with an opinionated LangChain-compatible adapter that promotes frequen","readme":"# AgentXin Plugin: Milvus Vector Store\n\n## Overview\n\nThis package exposes the Milvus vector store integration for the [AgentXinAI](https://github.com/agentxin-ai/agentxin) platform. It wraps the official `@zilliz/milvus2-sdk-node` client with an opinionated LangChain-compatible adapter that promotes frequently used metadata fields, handles hybrid-search ready schemas, and registers a server-side strategy via `@agentxin-ai/plugin-sdk`.\n\n## Key Features\n\n- Registers a global `VectorStoreStrategy` named `milvus`, ready to be consumed by the AgentXinAI agent runtime.\n- Provides an improved LangChain Milvus adapter that stores raw metadata safely while promoting filterable fields like `knowledgeId`, `documentId`, and `chunkId`.\n- Supports sanitized collection names and automatic partition provisioning for each knowledge base.\n- Enables configurable hybrid search, analyzer parameters, and credential management through environment variables.\n- Ships with lifecycle hooks (`onStart`, `onStop`) and structured logging for observability.\n\n## Installation\n\nTo use the plugin inside an AgentXinAI deployment:\nadd this plugin to the `PLUGINS` environment variable when starting the AgentXinAI system, and it will be loaded automatically:\n\n```ts\nPLUGINS=@agentxin-ai/plugin-dify\n```\n\n## Configuration\n\nThe plugin relies on NestJS `ConfigService` to resolve Milvus connection details. Set the following environment variables (or corresponding config entries) in your host application:\n\n| Variable | Description | Default |\n| -------- | ----------- | ------- |\n| `MILVUS_URI` | HTTP or gRPC endpoint of your Milvus instance, e.g. `http://localhost:19530` | `http://127.0.0.1:19530` |\n| `MILVUS_USER` | Username for Milvus authentication (if required) | `null` |\n| `MILVUS_PASSWORD` | Password for Milvus authentication (if required) | `null` |\n| `MILVUS_TOKEN` | Token-based auth string for Milvus Cloud or managed deployments | `null` |\n| `MILVUS_DATABASE` | Target database name | `default` |\n| `MILVUS_ENABLE_HYBRID_SEARCH` | Enables scalar + vector hybrid search features | `true` |\n| `MILVUS_ANALYZER_PARAMS` | JSON string describing analyzer params, e.g. `{\"type\":\"chinese\"}` | `null` |\n\n### Metadata Filtering\n\nThe adapter promotes several metadata fields (`enabled`, `knowledgeId`, `documentId`, `chunkId`, `parentChunkId`, `model`) for efficient filtering. Additional metadata is serialized into a JSON column, so arbitrary attributes remain queryable via hybrid search.\n\n### Deletion Helpers\n\nThe wrapped vector store overrides `delete()` to accept either LangChain-style filters or direct chunk ID lists. Internally it builds Milvus filter expressions like `chunk_id in [...]`.\n\n## Deployment\n\n- `npx nx release -p @agentxin-ai/plugin-milvus` runs package tests.\n- `npx nx run @agentxin-ai/plugin-milvus:nx-release-publish --access public --otp=<one-time-password-if-needed>` publishes to npm.\n\n## Requirements\n\n- Node.js 20+\n- Milvus 2.4+ (2.5+ recommended to leverage hybrid search)\n\n## Additional Resources\n\n- Milvus documentation: https://milvus.io/docs\n- AgentXinAI platform: https://agentxinai.cloud\n","readmeFilename":"README.md","_rev":"1-166ea3e696297dfa97a17d444b8882b0"}