Connect HeliosDB with your existing tools, frameworks, and platforms. Official SDKs, AI/ML framework integrations and wire-protocol compatibility let HeliosDB work alongside the tools you already use.
Client SDKs for Python, TypeScript, Rust and Go live in the open-source HeliosDB-SDKs repository. The Go module installs with go get; the Python, TypeScript and Rust SDKs are not yet published to PyPI, npm or crates.io, so you install them from source.
Client for Nano (branches and vector search; agent memory planned for Nano) and Lite (SQL and branches; vector and agent-memory endpoints planned), with LangChain and LlamaIndex integrations. Not on PyPI: install from source.
Typed client with a fluent query builder. Not on npm: clone the HeliosDB-SDKs repository, build sdks/typescript, then install the local package.
HTTP client for remote HeliosDB servers, not on crates.io yet: build it from the HeliosDB-SDKs source. For embedded use, add the published heliosdb-nano crate instead.
REST client for Nano (branches, vector search, time travel; agent memory planned for Nano) and Lite (SQL and branches; vector and agent-memory endpoints planned), fetched straight from the HeliosDB-SDKs repository.
Native integrations with the leading AI and machine learning frameworks. Use HeliosDB as your vector store, agent memory, and retrieval backend.
Vector store integration, chat message history, and retrieval chains. Drop-in replacement for any LangChain-compatible vector store.
Vector store index with metadata filtering, document retrieval, and hybrid search support for RAG pipelines.
Multi-agent framework integration with persistent tool state, shared memory, and conversation history storage.
Agent crews with persistent memory, shared knowledge bases, and task result caching across crew executions.
Code-Graph (hdb_code), Graph-RAG (WITH CONTEXT) and the native MCP server ship in HeliosDB Nano as build-from-source features; equivalents for Lite and Full are planned. Nano's MCP server answers over HTTP (POST /mcp, /mcp/ws, /mcp/sse); stdio clients use the HeliosDB CodeKB MCP binary, which embeds Nano.
Every edition serves the PostgreSQL wire protocol and a REST API, and Nano adds MySQL. On Nano, keep the HTTP and MySQL listeners on loopback (the default): API-key authentication for the HTTP API is planned, and a smaller listening surface is good practice anyway. HeliosDB Full opens listeners for PostgreSQL, MySQL, Oracle, MongoDB, Cassandra, Redis and SQL Server plus an HTTP gateway; MongoDB, Cassandra and SQL Server clients can connect and authenticate, and reading and writing HeliosDB tables through them is planned. Cards marked Planned are on the roadmap, not in a current release.
Full wire protocol compatibility. Use psql, pgAdmin, any PostgreSQL driver.
MySQL wire protocol support. Works with MySQL Workbench, Sequelize, and all MySQL drivers.
MongoDB wire listener in Full: mongosh, Compass and MongoDB drivers can connect and authenticate (SCRAM-SHA-256 only). Reading and writing HeliosDB tables through it is planned.
RESP listener in Full for redis-cli and Redis client libraries. Data written over RESP is held in memory only, not in HeliosDB tables.
CQL native-transport listener in Full. The DataStax drivers and cqlsh cannot connect to it yet; driver compatibility and reading and writing HeliosDB tables through CQL are planned.
ClickHouse native protocol for analytics workloads. Planned; there is no ClickHouse listener in the current Full release.
A GraphQL API for Lite and Full is planned. Today HeliosProxy includes an optional GraphQL gateway (flat top-level selections) that you enable in its configuration.
RESTful API with OpenAPI spec generation. Query, insert, and manage data over HTTP.
Snowflake wire protocol compatibility for warehouse migration. Planned; the Full HTTP gateway answers Snowflake routes with 501 Not Implemented today.
Oracle Net (TNS) protocol support for migrations from Oracle.
TDS listener in Full: SQL Server clients can connect and authenticate. Reading and writing HeliosDB tables through it is planned.
Pinecone-compatible REST API for vector search migration. Planned; the Full HTTP gateway answers Pinecone routes with 501 Not Implemented today, and Full serves vector search over its own HTTP vector-store API.
DRDA protocol support for IBM DB2 clients. Planned; there is no DRDA listener in the current Full release.
Deploy, manage, and connect HeliosDB with the tools and platforms your team already uses.
SQL intellisense, inline query runner, schema explorer, and result visualization directly in your editor.
A container image is published at ghcr.io/heliosdatabase/heliosdb-nano (tags v4.41.1 and latest); it contains the default-feature build, for amd64 and arm64. HeliosProxy is at ghcr.io/heliosdatabase/hdb-heliosdb-proxy. Lite and Full images are provided to licensed customers by HeliosDB sales.
Helm charts and Kubernetes operators for automated deployment, scaling, and lifecycle management.
Native database connector for building internal tools. Query HeliosDB directly from Retool dashboards and apps.
Workflow automation with triggers on data changes, new rows, and query results. Connect to 5,000+ apps.
Open-source workflow automation with HeliosDB nodes for reading, writing, and triggering on database events.
Extend your HeliosDB deployment with intelligent support and autonomous database administration.
AI-powered knowledge assistant trained on HeliosDB documentation. Instant answers with ready-to-run SQL. Free on Cloud, API key for on-prem.
Learn more →Autonomous database administration with proactive monitoring, self-healing, anomaly detection, and performance optimization. For Lite and Full.
Learn more →HeliosDB Full's HTTP gateway (default port 8443) includes an Agent Memory API for AI applications that need persistent conversation context: create sessions, store messages, search them and summarize long conversations over REST. An agent-memory REST API for HeliosDB Nano is planned (the current /v1/agents/memory routes are not functional yet); with Nano today, keep messages and their embeddings in a table and query them over the PostgreSQL wire (see the agent memory example). An agent-memory, vector-store and document-ingest REST API for Lite is planned.
# Create a session (HeliosDB Full)
POST /api/v1/agents/memory
{"title": "support-bot",
"metadata": {"user_id": "user-123"}}
# Add messages
POST /api/v1/agents/memory/:id/messages/batch
{"messages": [{"role": "user",
"content": "How do I reset my password?"}]}
# List the session's messages
GET /api/v1/agents/memory/:id/messages
# Search the session's messages
POST /api/v1/agents/memory/:id/search
{"query": "password reset", "top_k": 5}
# Summarize the most recent messages
POST /api/v1/agents/memory/:id/summarize
{"max_messages": 40}Get started with HeliosDB and connect it to your stack in minutes.