Where is it, and what calls it?
who_calls("billing.charge") → 3 callers, 2 repos
Context and memory for LLMs and agents
GraphRAG
Ground LLM answers in your knowledge graphAgentic AI
Agents grounded in how your domain connectsAgentic memory
Queryable memory that carries context across sessionsChatbots
Context-aware chatbots with long-term memoryProven patterns on connected data
Fraud detection
Trace rings of accounts, devices and transactions liveCybersecurity
Assets, findings and threats queried as one graphSupply chain
See how one supplier delay ripples downstreamReal-time recommendations
Suggestions based on how things actually relateAuryn plugs into the coding agent you already use. You ask in plain language; your agent answers from a graph of your whole codebase instead of reading its way through files.
ask a question in plain language
“how does checkout flow?”decides what to look up
trace_entrypoint(…)follows the real connections
CALLS · CALLS_APIholds the graph, on your machine
~5 ms lookupsone or many, every language
TS · C# · Python · SQLA real session on Meridian, a sample storefront made of four repositories: a TypeScript web portal, a C# API gateway, a Python platform core and a SQL database. Nothing to configure and nothing indexed upfront: the agent builds the graph on the first question and keeps using it.
Auryn connects itself to the coding agent you already use and keeps its graph on your machine. Nothing to configure, nothing added to your repository.
Restart your agent and ask your first question. A shared database for your team, a scripted install, or any other agent: see Docs → Connect your agent.
A small set of read-only tools, each answering one kind of question about your code.
who_calls("billing.charge") → 3 callers, 2 repos
find_path(SubmitOrderButton, ORDER_LINES) → 3 hops
impact_of_diff(HEAD~1) → 2 callers, 1 job, 1 screen
trace_entrypoint("POST /api/orders") → browser to table
Plus search by meaning, raw read-only Cypher and the tools that build the graph. The full list: Docs → Agent tools.
Every language gets a fast structural pass. Python and C# also get a full semantic analysis, so calls land on the code that actually runs. Install only what you need with extras, e.g. "falkordb-auryn[python,csharp]".
Tables, columns, keys, indexes and views from PostgreSQL, MySQL, SQLite and more.
Packages, procedures, functions and triggers, read straight from the database.
Every ingestor writes the same kinds of nodes and relationships, so one query follows a request from the browser, through the API, to the table it writes, regardless of the language at each layer.
Every node and edge, per language: Docs → Graph model.
Every command, for when you want to drive it yourself.
↗ GuidePut a whole system in one graph.
↗ GuideShare one graph across a team.
↗ GuideGate a merge on what a diff could break.
↗ ExtendAdd your own connections to the graph.
↗ TrustWhat Auryn touches, and how to remove it.
↗One command, about five minutes, entirely on your machine.