# Helix documentation

> Helix is Tray.ai's governed runtime for AI-built apps: developers and business builders create production TypeScript/React applications with AI coding agents (Claude Code, Cursor), deploy them with one command, and IT governs credentials, policies, budgets, and audit centrally.

Availability: Helix is pre-GA as of July 2026. Design Partner slots are open now and Early Access opens in the second half of August 2026; access starts with a conversation via https://helix.tray.ai.

Key facts (shipped as of August 2026): the CLI is `@trayai/helix-cli` (binary: `helix`, v0.85.0+, Node >= 24.18.0); the SDK is `@trayai/helix-sdk`; projects are configured in `helix.config.ts` (name, projectId, workspaceId, environment, auth aliases); you build by talking to Claude Code, which connects to the CLI's MCP server; `helix dev` runs locally and `helix deploy` ships to a managed runtime; deployed apps are TypeScript single-page apps served at `{project-id}.helix-app.ai` and always run BEHIND LOGIN, never public; the `ctx` object provides a project-scoped key-value store (`ctx.kv()`: getItem/setItem/hasItem/removeItem), logging (`ctx.log`), and authenticated third-party calls via aliases so credentials never enter code; the dashboard is at https://app.helix.tray.ai.

Features that are not in the product today are not documented here while they remain unshipped. Treat this index as closed: if a capability has no page below, Helix does not have it yet, whatever platform design material may describe. The authoritative index of what exists today is https://helix.tray.ai/documentation/reference/limits/.

Every page below is also available as raw markdown by appending .md to its URL. The complete corpus in one file: https://helix.tray.ai/documentation/llms-full.txt

## Getting started

- [Getting started](https://helix.tray.ai/documentation/getting-started.md): The path from an empty terminal to a deployed Helix app: what Helix is, the quickstart, building with Claude Code, and project anatomy.
- [What is Helix?](https://helix.tray.ai/documentation/getting-started/what-is-helix.md): Helix is Tray's platform where AI coding agents build production TypeScript and React apps, and IT governs everything they ship.
- [Quickstart](https://helix.tray.ai/documentation/getting-started/quickstart.md): Install the Helix CLI, log in, scaffold a project, build with Claude Code, and deploy your first app to Tray's cloud in one command.
- [Building with Claude Code](https://helix.tray.ai/documentation/getting-started/building-with-claude.md): How Claude Code works in a Helix project: the MCP server it needs, prompting patterns that get results, and when to run the CLI yourself.
- [Project structure](https://helix.tray.ai/documentation/getting-started/project-structure.md): What helix init scaffolds and why: helix.config.ts, file-based routing under functions/, the optional app/ front end, and .mcp.json.
- [Environment setup prompt](https://helix.tray.ai/documentation/getting-started/environment-setup.md): A copy-paste prompt that has Claude Code check your machine, install Node and the Helix CLI, and report back, without touching credentials.

## Core concepts

- [Concepts](https://helix.tray.ai/documentation/concepts.md): Four short pages that build a working mental model of Helix: architecture, workspace hierarchy, configuration, and the security model.
- [How Helix works](https://helix.tray.ai/documentation/concepts/how-helix-works.md): Where Helix code runs: a local Nitro dev server on localhost:3000, one AWS Lambda per project in production, and shared platform services.
- [Projects and workspaces](https://helix.tray.ai/documentation/concepts/projects-and-workspaces.md): How Helix is organized: organizations contain workspaces, workspaces contain projects, and permissions are granted at the workspace level.
- [The configuration model](https://helix.tray.ai/documentation/concepts/configuration.md): One helix.config.ts per project holds the name, project and workspace IDs, environment, and auth aliases. The CLI writes it for you.
- [Security model](https://helix.tray.ai/documentation/concepts/security-model.md): Why Helix code can call authenticated APIs safely: credentials stay on the platform, injected server-side, with every call audit-logged.

## IT & governance

- [IT & governance](https://helix.tray.ai/documentation/governance.md): How access works in Helix today: organization roles, workspace membership, per-app access control, credential aliases, and the org-wide activity trail.
- [Audit and compliance](https://helix.tray.ai/documentation/governance/audit-and-compliance.md): What Helix records: an org-wide activity trail of administrative actions, per-request execution traces, and the retention questions still open.
- [Identity administration](https://helix.tray.ai/documentation/governance/identity-administration.md): The three memberships that decide who can build and who can use a Helix app: organization role, workspace membership, and per-project access.
- [The Helix dashboard](https://helix.tray.ai/documentation/governance/dashboard.md): A tour of the builder dashboard: workspace switcher, project cards, and the Overview, Access Control, Deployments, and Logs tabs on every project.
- [The Admin console](https://helix.tray.ai/documentation/governance/admin-console.md): The org-wide view for Helix admins: every project, deployment, authentication, and administrative action across all workspaces, in one place.

## Guides

- [Guides](https://helix.tray.ai/documentation/guides.md): Task-focused guides for building on Helix: functions and routing, the key-value store, connected services, streaming, identity, and deployment.
- [Functions and routing](https://helix.tray.ai/documentation/guides/functions-and-routing.md): Define HTTP endpoints as files: the path sets the URL, the suffix sets the method, and Zod schemas validate input and output.
- [Frontend apps](https://helix.tray.ai/documentation/guides/frontend.md): Add a standard Vite and React app in app/, call your functions from the same origin with no CORS setup, and serve it from CloudFront in production.
- [Middleware](https://helix.tray.ai/documentation/guides/middleware.md): Run code before and after your handlers with functions/_middleware.ts files that nest from the project root down to each route.
- [Streaming](https://helix.tray.ai/documentation/guides/streaming.md): Return Web Streams responses from functions: Server-Sent Events, LLM pass-through, and NDJSON, with Lambda response streaming in production.
- [Key-value store](https://helix.tray.ai/documentation/guides/key-value-store.md): Store JSON values per project with ctx.kv(): getItem, setItem, hasItem, and removeItem. In memory locally, persistent once deployed.
- [Connected services](https://helix.tray.ai/documentation/guides/connected-services.md): Call third-party APIs with ctx.http and auth aliases. Helix injects credentials server-side, so tokens never appear in your code or logs.
- [Identity and roles](https://helix.tray.ai/documentation/guides/identity-and-roles.md): Published Helix apps are always behind login. Control who can build in a workspace and who can open the app from its Access Control tab.
- [Deployment](https://helix.tray.ai/documentation/guides/deployment.md): Build and deploy in one command. Functions run on Lambda, SPA assets on the CDN, plus schema migrations, secrets, and rolling back.
- [Observability](https://helix.tray.ai/documentation/guides/observability.md): Write structured logs with ctx.log, then read one traced entry per HTTP request in the Helix dashboard. Local runs pretty-print to your terminal.

## Reference

- [Reference](https://helix.tray.ai/documentation/reference.md): Exact commands, flags, configuration keys, and context APIs for Helix, organized for lookup rather than learning.
- [Helix CLI](https://helix.tray.ai/documentation/reference/cli.md): Install the Helix CLI and look up every shipped helix command, from login and init through dev, auth, deploy, and MCP.
- [Project and auth session commands](https://helix.tray.ai/documentation/reference/cli/project.md): Log in to Tray, scaffold a project with helix init, choose a workspace, and point the CLI at a project ID and environment.
- [Local development commands](https://helix.tray.ai/documentation/reference/cli/dev.md): Run your project locally with helix dev, pick a port and a log format, and see how local storage and auth differ from production.
- [Authentication commands](https://helix.tray.ai/documentation/reference/cli/auth.md): Connect a service, map an alias to an existing authentication by UUID, and list the aliases your Helix project uses.
- [Deploy commands](https://helix.tray.ai/documentation/reference/cli/deploy.md): Deploy a project with helix deploy, see what the first deploy provisions, and check a deployment's status and build errors.
- [helix.config.ts reference](https://helix.tray.ai/documentation/reference/configuration.md): Every key helix.config.ts accepts today: name, projectId, workspaceId, environment, and authentications, with types and semantics.
- [The ctx object](https://helix.tray.ai/documentation/reference/context.md): Field-by-field reference for the ctx object Helix functions receive, plus getContext() for shared code and availability by function type.
- [ctx.kv()](https://helix.tray.ai/documentation/reference/context/kv.md): Reference for ctx.kv(), covering scope keywords, external aliases, and the full key-value API including getItem, setItem with TTL, getKeys, and incr.
- [ctx.http and ctx.auth()](https://helix.tray.ai/documentation/reference/context/http.md): Reference for ctx.http, the ky HTTP client, the authed() proxy rewrite that injects credentials server side, and ctx.auth() for non-sensitive variables.
- [ctx.log](https://helix.tray.ai/documentation/reference/context/log.md): Reference for ctx.log with the info, warn, error, and debug signatures, metadata serialization, trace correlation, and dev versus production transport.
- [ctx.identity](https://helix.tray.ai/documentation/reference/context/identity.md): Reference for ctx.identity with the user and org shape, roles as the authorization source, null on excluded paths, and dev identity injection.
- [Request and response](https://helix.tray.ai/documentation/reference/context/request.md): Reference for request and response on ctx, covering validated input, headers, raw body access, status codes, errors, middleware state, and triggers.
- [Function definition reference](https://helix.tray.ai/documentation/reference/functions.md): Exact signatures for defineFunction, defineSchedule, defineQueueConsumer, defineAppTrigger, and defineMiddleware, plus the named exports the runtime reads.
- [Limits, defaults, and roadmap](https://helix.tray.ai/documentation/reference/limits.md): Runtime limits and defaults for Helix functions, storage, queues, and AI calls, plus one list of every feature that is planned but not yet shipped.

## Troubleshooting

- [Troubleshooting](https://helix.tray.ai/documentation/troubleshooting.md): Fixes for common Helix failures: install and PATH problems, Node and CLI versions, deploy and build errors, local key-value resets, and access surprises.

## Optional

- [Helix website](https://helix.tray.ai): product overview and platform pages
- [Helix dashboard](https://app.helix.tray.ai): manage projects, access, deployments, and execution logs