Documentation Index¶

Fetch the complete documentation index at: https://makakoo-traylinx-35.mintlify.app/llms.txt Use this file to discover all available pages before exploring further.

Studio Tools¶

Build, configure, and deploy AI tools and agents using TrayLinx's Studio Tools.

Studio Tools is TrayLinx's creation environment for AI tools and agents. Use it to build custom AI assets — from lightweight agents to full MCP servers — configure their endpoints and deployment settings, manage versions, and publish them to the Marketplace.

What are Studio Tools?¶

A Studio Tool is an asset you own and manage inside a project. Studio Tools have a type, a unique entity ID, version history, deployment configuration, and optionally an exposed HTTP endpoint. Once created, you can install tools from other projects or the Marketplace alongside your own.

Asset types you can create

API type	User-facing label	Subtypes (AI only)
`ai`	AI & Automation	`agent`, `mcp_server`, `prompt_template`, `ml_model`
`dataset`	Dataset	—
`app`	Application	—
`media`	Media	—
`defi`	DeFi & Finance	—
`security`	Security & Identity	(Sentinel Pass via guided flow)

Creating a new tool¶

TrayLinx offers two creation paths: Express Agent for a fast single-step setup, and the modular dialog for full control over all fields.

Express Agent¶

Use Express Agent when you want to create an AI agent quickly with sensible defaults.

Navigate to your project and click the AI Studio tab.

Click Express Agent in the toolbar or empty state.

Provide:

* **Agent name** (required)
* **Description** (optional)
* **Model** (required) — choose from available models such as GPT-4, Claude 3, or Gemini Pro

Click Create. The agent is created with default settings: temperature 0.7, max tokens 2000, memory enabled, and private visibility.

Modular creation dialog¶

Use the modular dialog when you need to set specific types, tags, visibility, or asset-level metadata.

Click New Tool (or New Asset) in the Studio Tools toolbar.

Choose from AI & Automation, Dataset, Application, Media, DeFi & Finance, or Security & Identity.

If you selected AI & Automation, choose a subtype:

* **Agent** — a conversational AI agent
* **MCP Server** — a Model Context Protocol server
* **Prompt Template** — a reusable prompt
* **ML Model** — a machine learning model asset

Enter:

* **Title** — 1–100 characters, letters, numbers, and common punctuation
* **Entity ID** — lowercase letters, numbers, hyphens, and underscores (max 50 characters); used as a stable identifier
* **Description** — up to 1000 characters
* **Visibility** — `publicly_visible` or `privately_visible`
* **Tags** — up to 10 tags, each up to 30 lowercase characters

Click Create. The tool appears in your Studio Tools list.

Selecting Security & Identity routes you to the Sentinel Pass creation flow, which uses an agent-based pipeline to generate the pass.

Configuration options¶

Deployment configuration¶

AI tools and applications can be deployed as live endpoints. Open the Deployment Config dialog from the tool's action menu to set:

Environment — development or another target environment
Resources — optional CPU, memory, and replica settings
Environment variables — key=value pairs, one per line

Deployment moves through the states: not_deployed → pending_deployment → deploying → deployed. A deployment_failed state indicates an error; retry or review the configuration.

Endpoint configuration¶

Once deployed, a tool exposes an HTTP endpoint. View and update endpoint settings from the tool detail view. The endpoint URL is the address other services or the Personal Assistant can call to invoke the tool.

Endpoint testing¶

Use the built-in Endpoint Tester to validate your tool's endpoint without leaving TrayLinx.

Navigate to a deployed tool and open the Endpoint Tester panel.

Set:

* **HTTP method** — GET, POST, PUT, DELETE, PATCH, HEAD, or OPTIONS
* **Headers** — JSON object of request headers
* **Body** — request payload (for POST/PUT/PATCH)
* **Timeout** — request timeout in milliseconds (default 10 000 ms)

Click Test. The tester shows the response status, body, and response time.

The tester stores a history of past test runs so you can compare responses across changes. Advanced settings include redirect following, SSL validation, and retry configuration.

Version management¶

Every Studio Tool you own supports version tracking. Open the Version Management dialog from the tool's action menu.

For owned tools: Click Increment Version to bump the tool's version number. Add optional release notes to describe the changes.

For installed tools: Use Upgrade or Downgrade to switch the installed version to a different available release. An impact analysis shows what will change before you confirm.

Increment your tool's version before publishing a new release to the Marketplace so consumers can pin to stable versions.

Installing vs. creating tools¶

You have two ways to get a tool into your project:

Build your own tool from scratch using the modular dialog or Express Agent. You own and manage the tool; it lives under your project.

Browse the Marketplace and install a tool created by someone else. The tool runs at its published version and can be upgraded when new versions are released.

Installed tools appear in your project's tool list alongside owned tools. You can configure them independently, but you cannot edit their core definition.

Publishing to the Marketplace¶

If you have built a tool or agent you want to share, you can publish it to the TrayLinx Marketplace directly from Studio Tools.

From the tool's action menu, select Publish to Marketplace (or use the AIStudioMarketplace panel inside AI Studio).

Provide:

* **Title** — the display name on the Marketplace
* **Description** — what the tool does
* **Category** — for example, `productivity`, `developer_tools`, `personal_assistant`, `creative`, `business`, `finance`, `health`, `travel`, or `games`
* **Pricing model** — `free`, `usage_based`, or `tiered`
* **Price** (usage-based only) — price in cents per use

Click Publish. Your tool becomes visible to other TrayLinx users in the Marketplace.

To remove a listing, use Unpublish from the same menu. The tool remains in your project but is no longer discoverable in the Marketplace.

Built with Mintlify.