OYNIX
Integrations

Bring your own keys

Start free. Add a key when you want more.

Oynix runs entirely on your machine out of the box — no keys required. Each key you add simply switches on another capability. Your keys and your database stay yours (BYOK + BYODB).

Works out of the box · no keys

With just the CLI and a local model, you already get:

  • Index your whole codebase into a live knowledge graph
  • Ask the graph in plain English — and “why does this exist?”
  • Auto-generated docstrings and team skills
  • Semantic search — find code by meaning, not just keywords
  • Presence & conflict detection across your team

Add a key, get more

Everything above keeps working. These are optional — add a key only for what you want.

Want faster, sharper answers (and team scale)?

Add: A cloud model key — Anthropic, OpenAI, Gemini, Groq or OpenRouter

Swaps the local model for a cloud one: quicker, higher-quality answers that scale to a whole team. Everything above keeps working — it just gets better.

Want it to read images & diagrams in docs?

Add: A multimodal key — Google Gemini or OpenAI (or a local vision model)

Oynix reads diagrams, screenshots and images inside your documents and folds them into the graph.

Want meetings & voice notes searchable?

Add: Speech-to-text — runs locally, no key needed

Meetings, voice notes and recordings get transcribed and become part of your searchable knowledge.

Want even sharper search?

Add: An embeddings key — Voyage or OpenAI

Higher-quality meaning-matching than the free local default, so retrieval gets more precise.

How to add a key

Add a provider key whenever you want to switch on its capability — the CLI picks it up automatically.

$ oynix config set anthropic.api_key sk-...

Prefer environment variables? ANTHROPIC_API_KEY, OPENAI_API_KEY, GEMINI_API_KEY, GROQ_API_KEY and VOYAGE_API_KEY all work too.

Bring your own database

Everything Oynix builds is stored in a graph database you own — not ours. It needs a Neo4j database; here’s exactly what to give it and how it works.

FieldRequiredWhat it isExample
NEO4J_URIRequiredWhere your database lives — the connection address Oynix dials.neo4j+s://xxxx.databases.neo4j.io · bolt://localhost:7687
NEO4J_USERNAMERequiredThe database user. Almost always “neo4j”.neo4j
NEO4J_PASSWORDRequiredThe password for that user — from Aura, or one you set yourself.••••••••
NEO4J_DATABASEOptionalWhich database to use. Only needed if you renamed it from the default.neo4j (default)
01

Get a graph database

Oynix keeps your entire knowledge graph — code, decisions, docs and the links between them — in a Neo4j database that you own. Pick whichever is easier:

  • Easiest (cloud): create a free database at neo4j.com/aura. It hands you a connection URI and a password — that's everything you need.
  • Self-hosted: run Neo4j yourself (one Docker command), or point at an instance you already have.
  • Either way it's your database — Oynix never sees or stores your data anywhere else.
02

Give Oynix the connection details

Oynix needs four things to connect — three are required. Set them once with the CLI, or as environment variables (the table above explains exactly what each one is).

$ oynix config set neo4j.uri neo4j+s://xxxx.databases.neo4j.io
  • Then set the username and password the same way: oynix config set neo4j.username neo4j and oynix config set neo4j.password ••••.
  • Prefer environment variables? Export NEO4J_URI, NEO4J_USERNAME and NEO4J_PASSWORD instead — same effect.
  • You only need NEO4J_DATABASE if you renamed your database away from the default.
03

Start the engine

Bring everything up. Oynix connects to your database, sets up what it needs on the first run, and from then on every read and write goes to your instance — nowhere else.

$ oynix mcp start
  • First run creates the indexes automatically — nothing to configure by hand.
  • Confirm it connected anytime with oynix status.