Syncanix documentation
Everything you need to discover your API, embed the chat widget, connect an MCP client, and ship an AI assistant that takes real actions for your users — hosted in the EU, in six languages.
Introduction
- What is Syncanix?Syncanix turns your existing API into an AI assistant your users can talk to — embedded in your app and available to AI clients over MCP.
- How Syncanix worksFrom one discovery command to a live assistant: how the capability catalog connects the CLI, the widget, the chat surface, and the MCP server.
- Architecture overviewHow Syncanix connects your users, your API, and the AI — and where it all runs.
- Languages & RTLSyncanix works in six languages out of the box, with full right-to-left support for Arabic and Hebrew.
- GlossaryKey terms used across the Syncanix documentation.
Getting started
Core concepts
- The capability catalogThe structured list of what your API can do — discovered once, authored by you, and read by every surface.
- The browser is the gateSyncanix’s security model in one idea: the AI never holds your users' credentials.
- Identity & authenticationSyncanix uses your existing identity provider — it never becomes your users' password store.
- Action confirmation & riskHow Syncanix decides what runs automatically and what needs a human yes.
- Rendering modesHow the assistant shows results — built-in components, your own components, or sending the user to the right page.
- Reliability & model routingHow Syncanix keeps answering: multiple model providers, your own keys, and EU routing.
- Memory & knowledgeWhat the assistant remembers across sessions, and the knowledge it draws on to answer.
CLI
- The Syncanix CLIDiscover your API and manage your catalog from the command line — no install required.
- syncanix initDiscover your API, build the capability catalog, and upload it — with consent-gated AI enrichment.
- More CLI commandsSign in, import a schema, run the local harness, diagnose, register an external MCP server, and apply catalog overlays.
- Configuration filesThe files the CLI reads and writes: ignore rules, CLI config, the catalog, your consent record, credentials, and the catalog overlay.
- Environment variables & exit codesThe environment variables the CLI reads and the exit codes it returns.
- Framework supportThe languages and frameworks the discovery CLI can read automatically — and what to do when yours is not one of them.
Framework guides
- Syncanix with ExpressMake an Express API agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with NestJSMake a NestJS API agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with Next.jsMake a Next.js app agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with FastifyMake a Fastify API agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with FastAPIMake a FastAPI service agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with DjangoMake a Django project agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with RailsMake a Rails API agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
- Syncanix with LaravelMake a Laravel API agent-ready: what discovery reads, how to run it, and how to ship the chat surface and MCP server.
Dashboard
- The dashboardWhere you configure your workspace — the catalog, persona, knowledge, design, team, and billing — all in one place.
- Catalog & capabilitiesBrowse the discovered capabilities and refine each one — its description, permissions, risk, and more.
- Custom endpoints & tool availabilityAdd capabilities the assistant can call beyond your discovered routes, and control which users each tool is offered to.
- Prompts & personaGive the assistant a voice, a system prompt, and clear fallback and refusal behaviour.
- Knowledge & docsConnect content sources so the assistant can answer from your documentation, and tune what it retrieves.
- Design & the Widget Design StudioStyle the widget to match your brand — theme tokens, components, and the per-tenant design studio.
- Monitoring, cost & auditSee how the assistant is performing, what it costs, and a full record of what happened.
- Environments & promotionSet things up safely in development, then promote to production behind an evaluation gate.
- Team & rolesInvite your team and give each person the right level of access with five clear roles.
- SettingsKeys, origins, your own model keys, identity federation, limits, billing, and discounts.
Widget
- The widgetEmbed the Syncanix chat widget in your app with a single script tag — isolated, lazy, and on-brand.
- Widget configurationConfigure the widget with data-* attributes on the script tag.
- Widget runtime APIControl the widget from JavaScript through the window.syncanix object.
- Theming the widgetThe widget matches your brand automatically, with full control when you want it.
- Sizes & modesThe widget comes in five sizes and two layout modes — pick what fits each place you use it.
- The chat experienceWhat your users see in a conversation — streaming replies, suggestions, action cards, disclosures, and more.
- The headless SDKBuild a completely custom chat UI on top of Syncanix, without mounting the widget.
- Accessibility & RTLThe widget meets WCAG 2.1 AA and lays out correctly in right-to-left languages.
MCP server
- The MCP serverEach workspace gets its own MCP server, so AI clients like Claude Desktop, Cursor, and ChatGPT can reach your capabilities.
- Connect a clientConnect Claude Desktop, Cursor, ChatGPT, and other MCP clients to your workspace.
- How MCP sign-in worksMCP clients authenticate with OAuth 2.1, federated to your identity provider — Syncanix never holds your users' credentials.
- Tools & executionWhat the MCP server exposes, how risky tools are marked, and how calls actually run.
- MCP limits & endpointsSession limits and the OAuth and transport endpoints the MCP server exposes.
Authentication & access
- Authentication & accessHow people sign in to Syncanix — your team to the dashboard, your own users to the chat — plus the keys, identity providers, and safeguards that connect them.
- Signing in & workspacesHow you and your team sign in to the Syncanix dashboard, and how workspaces keep each account’s work cleanly separated.
- Authenticating your usersThe four ways the widget can authenticate your own end users, and how to choose the one that fits your stack.
- Connecting an identity providerConnect Auth0, Clerk, Amazon Cognito, or WorkOS in the dashboard so Syncanix can authenticate your end users through the provider you already use.
- API keysCreate, reveal, and revoke the two kinds of key Syncanix issues — a secret key for the CLI and MCP, and a publishable key for the embedded widget.
- Step-up & action safetyHow Syncanix classifies every action as read, write, or destructive — and the confirmation, re-verification, and undo controls that protect each one.
Billing & plans
- Billing & plansHow Syncanix charges: per-workspace plans, end-user metering, subscriptions, bringing your own model key, and the limits that keep usage predictable.
- PlansThe five Syncanix plans — Dev, Starter, Growth, Scale, and Enterprise — plus trials, annual billing, and discounts for nonprofits, education, and open source.
- Usage & meteringHow Syncanix meters usage with monthly active end users (MAEU), how the count works, what happens at your limit, and where to track it.
- Managing your subscriptionStart, change, or cancel a Syncanix plan, manage your payment method through the Paddle billing portal, and find your invoices and receipts.
- Bring your own key (BYOK)Use your own Anthropic or OpenAI key so model usage is billed directly by your provider with no Syncanix markup. Keys are write-only and encrypted at rest.
- Rate limits & spend capsThe guardrails that keep Syncanix usage predictable: platform rate limits that protect stability, and spend caps you configure to stay within budget.
Trust, privacy & compliance
- Trust, privacy & complianceHow Syncanix protects your data and your end users’ data: EU data residency, tenant isolation, what we process and retain, your privacy rights, our compliance program, and AI transparency.
- Data residency & securitySyncanix runs entirely in AWS eu-central-1 (Frankfurt), isolates each customer’s data at four layers, and encrypts data in transit and at rest.
- Data handling & retentionWhat data Syncanix processes, how zero-data-retention (ZDR) works with model providers, what the discovery CLI sends, and how long data is kept.
- Privacy & your rightsHow assistant memory is controlled, how personal data is handled, and how you and your end users can exercise GDPR and CCPA rights through the data request form.
- Compliance & subprocessorsSyncanix’s compliance program: GDPR and CCPA support with a DPA available, SOC 2 and ISO 27001 in progress, and the subprocessors we rely on.
- AI transparencyHow Syncanix discloses AI use to end users in every language, which models power the assistant, and how answers are grounded in your content.
Guides
- GuidesTask-oriented walkthroughs that combine the reference docs into complete jobs — launching the widget, connecting an AI client, grounding answers, acting for a user, and going live.
- Add the chat widget to your siteGo from your codebase to a live chat widget: discover your API with the CLI, review the catalog, design the widget, and embed the snippet.
- Connect an AI client to your workspaceConnect Claude Desktop, Cursor, or ChatGPT to your Syncanix MCP server so an AI client can use your API on your behalf.
- Ground answers in your contentAdd your documentation and content so the assistant answers from your material instead of guessing, and check that it’s working.
- Let the assistant act for a signed-in userAllow the assistant to take authenticated actions on behalf of a signed-in end user, scoped to their permissions, with confirmation before sensitive actions.
- Keeping your catalog freshRe-run discovery when your API changes, wire it into CI, and let runtime drift detection catch what slipped through.
- Data export and erasureHow data-subject requests work with Syncanix — submitting access, export, and erasure requests, the response windows, and who does what.
- Go to productionMove your assistant from test to live: switch to the production environment, mint live keys, set allowed origins, choose a plan, and turn on guardrails.
- TroubleshootingSymptom-first fixes for the issues we actually see: a blocked widget, a slow first request, detection misses, enrichment failures, and sign-in problems.
Reference
- ReferenceA consolidated lookup: quick facts, key and ID formats, platform limits and defaults, and pointers to the CLI reference, MCP reference, framework support, and glossary.
- Keys, tokens & IDsA consolidated lookup of the credential and identifier formats used across Syncanix: secret and publishable API keys, MCP connection tokens, and environments.
- Limits & defaultsA single lookup for cross-cutting platform facts: data region, log retention, rate-limit responses, the metering unit, spend caps, and the supported languages.
- Runtime API referenceThe customer-side runtime contracts: the signed intent header your backend verifies, the act-as-user contract it observes, and the witness payload shape.