Welcome to Nolixan
The governed-execution control plane for production AI agents.
Nolixan lets your AI agents safely act in the real world — sending the Slack message, opening the GitHub PR, charging the invoice — with policy, approvals, and a tamper-evident audit trail wrapped around every call.
You bring an agent; Nolixan gives it governed hands.
What you get
- Connections — OAuth and API-key auth to hundreds of providers, with credentials encrypted at rest (AES-256-GCM) and refreshed for you. Connecting accounts →
- Actions & tools — a typed catalog of provider actions (
provider__action), callable directly, over MCP, or as native tools in your agent framework. - Governance — a fail-closed policy gate (default-deny on writes, plus a
sanitize-and-allow
transformverdict), approvals with separation of duties, a hash-chained audit log, and signed, offline-verifiable receipts — the same controls enforced identically across API, SDK, CLI, MCP, triggers, and proxy. - Fleet & identity — per-agent DID identity, shadow-agent discovery, a live decision feed, spend budgets, SLOs, and a compliance (OWASP/SOC2) map. Governance you can prove, not just claim.
How it fits together
Your agent calls a tool → Nolixan resolves the connection and credentials → the policy gate decides allow / deny / needs-approval → the action runs against the provider → the result and the full decision are written to the tamper-evident audit log. The same path applies whether you call an action directly, through the proxy, or over MCP.
Start here
- Quickstart — connect a provider and run a governed action in a few minutes.
- Agent frameworks — drop Nolixan tools into OpenAI, Anthropic, Gemini, LangChain, CrewAI, or the Vercel AI SDK.
- Governance overview — why governed execution, and how the control plane works.
SDKs
Nolixan ships first-party SDKs for Node, Python, Go, and .NET. Pick your language in the Quickstart.