AI Development / Tray Headless
Build agents, tools, and integrations.
In your coding assistant. Fully governed.
Tray Headless turns Claude Code, Codex, and any coding assistant into a full builder: create agents, MCP tools, automations, and integrations on the full platform, with governance, audit, and access control built in. Dedicated plugins for Claude Code and Codex; MCP for Cursor, Windsurf, and any MCP-compatible client.
Purpose-built plugin
Tray Headless
for Claude Code
The deepest coding-assistant experience for Claude
Purpose-built plugin
Tray Headless
for Codex
The deepest coding-assistant experience for Codex
Any MCP-compatible coding assistant
Other coding assistants
Cursor · Windsurf · Gemini · and more
Platform-neutral
Tray Headless MCP
The full Tray platform as MCP services — built for developers to unlock its complete power from any coding assistant
The unified platform underneath
Tray AI Orchestration Platform
What it does.
Build the full range
Agents, MCP tools, automations, integrations. Not a subset, not a wrapper: the complete platform.
Governed by default
Everything you build inherits Tray's governance, audit trails, and access control. Move at coding-assistant speed without giving up enterprise control.
Native plugins
Dedicated plugins for Claude Code and Codex: config, typed tools, response formatters, validation hooks, and six built-in skills — ready out of the box.
MCP for any coding assistant
Tray Headless MCP exposes the full platform to Cursor, Windsurf, and any MCP-compatible client. Four capability groups: Build, Connect, Validate, Configure.
Validate as you build
Per-step schema checks and whole-workflow audits run as tool calls — catch misconfigured steps and missing fields before they reach production.
One platform underneath
700+ connectors, one execution engine, one governance model — the same Tray platform that runs 150B+ integrations a year. You build on all of it from your coding assistant.
Build it in your coding assistant. Govern it by default.
Tray Headless is the full platform, exposed to the coding assistant your team already uses. Whatever you build is fully governed: policy, audit trails, and access control by default.
- Agents, MCP tools, automations, and integrations — all from one place
- Governance, audit, and access control on everything you build
- Dedicated plugins for Claude Code and Codex, with CDK, skills, and validation
- MCP for Cursor, Windsurf, and any MCP-compatible coding assistant
- 700+ connectors and one execution engine underneath
What you can build
See what teams build with Tray Headless.
-
Build a governed agent from your coding assistant.
Describe the agent in Claude Code or Codex; Tray Headless scaffolds it, wires up the connectors and auth, with governance and audit in place from the first run.
-
Stand up an MCP tool, safely.
Expose a governed capability as an MCP tool without hand-rolling a server. Access control and observability come from the platform, not a bespoke setup.
-
Generate a connector from an API spec.
Point Claude CDK at an OpenAPI spec and get a production-grade Tray connector — typed inputs, output shapes, error handling — in minutes.
-
Scaffold and validate an automation.
Use the build-workflow skill to create a multi-step automation from a natural-language description, validated with per-step and whole-workflow checks before it reaches production.
-
Migrate off a legacy tool, fast.
One enterprise rebuilt 7 workflows in 30 minutes with Claude Code and Tray Headless, versus about two days by hand: 95% faster, with auto-generated docs and a full audit trail.
-
Build headless, manage visually.
Build and iterate in your coding assistant, then hand off to the ops team in Tray Build's visual canvas: same workflow, same governance, no migration.
Demo
See Tray Headless in action.
“Tray Headless will change the way we build workflows. I'm super excited about that, because we're trying to get ourselves to 10x and 100x of what we were doing last quarter, and Tray Headless will enable us to do that.”
Introducing Tray Headless
Frequently asked questions
What can I actually build with Tray Headless? +
The full range Tray supports: agents, MCP tools, automations, and integrations. It's the complete platform exposed to your coding assistant — not a limited subset — and everything you build carries the same governance and controls.
Which coding assistants does Tray Headless support? +
Any MCP-compatible coding assistant, including Cursor and Windsurf. Claude Code and Codex go further: dedicated plugins with config files, typed tool definitions, response formatters, validation hooks, and six built-in skills, wired up out of the box.
How does governance work when people build from a coding assistant? +
Everything built through Tray Headless inherits the same governance, audit trails, and access control as anything else on the platform. Building fast from a coding assistant doesn't mean building ungoverned.
What are the six built-in skills for Claude Code and Codex? +
build-workflow, tray-connectors, tray-patterns, tray-gotchas, research-connector, and set-workspace. Together they give Claude Code and Codex structured, validated access to the full Tray platform without manual MCP configuration.
Is Tray Headless MCP-only? +
No. Headless Core — Tray's full RESTful platform APIs — is available for teams who want direct API access without MCP. MCP is the recommended path for coding assistants; Headless Core keeps the platform open regardless of tooling.
How does Tray Headless relate to Tray Build? +
They're two interfaces to the same platform. Build headless in your coding assistant, and an operator can open the same workflow in Tray Build's visual canvas. Same connectors, same governance, same execution engine — different surface.
One platform
Tray Build and Tray Headless. Same platform.
Two first-class build experiences on the same connectors, governance layer, and execution engine. Not a tradeoff.
See Tray Headless in action.
Walk through a scenario from your stack with a Tray.ai expert.