Raycast

How to give Raycast access to ML Experiment Costing

Set up in minutes, once. From then on, Raycast gets things done for you: ML Experiment Costing becomes an HTTP MCP server, ready in the Action Panel.

Ask AI about this guide

ML Experiment Costing Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_LpCrDK7YbLXumDZ1Yg3SufrVlCdTOf49sRuTis94/mcp

Connecting ML Experiment Costing to Raycast

A few minutes, in Raycast Pro

  1. 1Search for the Install MCP Server command (or open Manage MCP Servers and choose Install New Server).
  2. 2Give the server a name and icon, select HTTP as the transport, and paste your ML Experiment Costing link as the URL. The link above is the URL.
  3. 3Press Install MCP Server and confirm the status shows Running, then ask something real. ML Experiment Costing is now part of Raycast.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "ml-experiment-tracking-cost-analyzer-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_LpCrDK7YbLXumDZ1Yg3SufrVlCdTOf49sRuTis94/mcp"
    }
  }
}

Raycast + ML Experiment Costing

One chat. Your ML Experiment Costing, live inside Raycast.

Real prompts, answered with live data. This is the kind of conversation you will have in Raycast once the server is in.

Waiting for input…

Works with modern AI clients that support MCP, including ChatGPT, Claude, Cursor, and more.

ChatGPTClaudeCursorPerplexityGeminiMicrosoft CopilotRaycastMeta AI

These prompts ship with the ML Experiment Costing Connector. Once yours is connected, they work in your sessions.

The full capability set, the prompts and the observed latency live on the ML Experiment Costing Connector page.

Mission Control

See everything your Raycast agents do in ML Experiment Costing.

When Raycast runs a tool in ML Experiment Costing, it happens on Raycast's side: you see the result, but not the request itself, what data it carried, or whether a rule was broken.

Now every move lands on your screen.

Every execution is recorded, enforced, and shown to you: what ran, what it returned, what was blocked, and why.

AI Briefing
Upgrade to unlock

Know exactly what needs your attention before you even look at the charts. One briefing. The full picture. The next move.

Requests
0
Total in this period
Avg Latency
0
API 88ms · Overhead 54ms
Vinkius Reliability
0
Agent 280 · Upstream 47 · Vinkius 15
Tokens
0
AI tokens processed
DLP Protected
0
Sensitive data redacted
Cost Saved
0
FinOps truncations applied

AI Agents Activities

last 30 days
30d agotoday

Request Volume & Latency

requests latency

Every request your Raycast agents make through ML Experiment Costing is logged, enforced and auditable. See how AI Governance turns agent activity into accountability, and what it means for your ML Experiment Costing workflows.

What sets us apart

It's not just the Connector that stays safe. Everything your agents do through it is.

Connecting ML Experiment Costing to your AI is step one. But an open line between your agents and ML Experiment Costing is only half the story: without protection, every action they take runs unguarded.

That's why protection doesn't stop at the door.

Every action your agents take through the Connector runs inside our infrastructure: signed activity logs, automatic key rotation, spending limits that pause instead of surprise, and a sandbox that isolates every execution.

01

Tamper Proof Activity Log

Every request your agents make in ML Experiment Costing is recorded and signed, so any change to the history becomes detectable.

02

Automatic Key Rotation

We rotate security keys every 24 hours. Your ML Experiment Costing connection stays protected without any manual work.

03

Spending Protection

We enforce the spending limit you choose and pause actions that exceed it until approval is received. Your ML Experiment Costing actions never surprise you.

04

Isolated Sandbox

Every execution through ML Experiment Costing runs isolated, with 34+ security rules covering memory, CPU, network access, file handling and execution.

05

Security Event Streaming

Security events flow to Splunk, Datadog or your own webhook, so ML Experiment Costing activity lands in the tools you already rely on.

06

Instant Resume

Idle connections resume in 3 to 5 ms while preserving their state, keeping your Raycast sessions responsive.

07

Malicious File Protection

We detect compressed files designed to exhaust system resources and block them before they ever reach ML Experiment Costing.

08

Credential Validation

We test your ML Experiment Costing credentials before saving them. Invalid credentials are rejected instead of being stored.

09

Separate Usage Limits

Usage stays isolated per account, so activity in one connection never consumes another account's allowance.

One Connector, one link, and the whole operation is protected. Not just the connection: everything your agents do through it.

These guarantees are not add-ons: they are the infrastructure every ML Experiment Costing connection runs on. Read how our infrastructure protects everything your Raycast agents do, end to end.

Inside Vinkius

Get ML Experiment Costing ready for your AI. It's easy.

You choose ML Experiment Costing from the Vinkius Catalog and get one connection link. That's it.

  1. You choose ML Experiment Costing

    You find ML Experiment Costing in the Catalog and install the Connector with one click.

    ML Experiment Costing

    Active

    Enable Connector
    Activating…

  2. You add the link in Raycast

    Vinkius gives you one connection link. In Raycast Pro, you install an HTTP MCP server with the link as the URL, and the tools are pulled in automatically.

    Token generated successfully

    MCP Connection URL

    https://edge.vinkius.com/vk_preview_LpCrDK7YbLXumDZ1Yg3SufrVlCdTOf49sRuTis94/mcp

    Connection Token

    vk_live_••••••••
    Go to Dashboard

Set it up once. Use ML Experiment Costing with the AI you already use.

FAQ

Raycast access questions.

  • 01

    Do I need Raycast Pro?

    Yes. Custom MCP servers are a Pro-exclusive feature of Raycast, and the same server works on your Mac and on iOS (which is HTTP-only).

  • 02

    If I connect ML Experiment Costing to Raycast, can I use the same connection in Claude or ChatGPT later?

    Yes, and that's the point: the connection is yours, not Raycast's. You paste the same link in any AI you use, and there's nothing new to set up or pay again. Connect a new AI and ML Experiment Costing is already there.

  • 03

    What does Raycast need to connect to ML Experiment Costing?

    Just the link, with HTTP as the transport. OAuth is supported if you prefer it, but the hosted Connector handles access on our side, so No auth works too.

  • 04

    What can I actually ask Raycast to do with ML Experiment Costing?

    Whatever the Connector covers: check data, run actions and bring answers back. The Connector page lists every capability your AI gains, with real examples you can copy.

  • 05

    Raycast doesn't show the Connector. What do I check?

    Open Manage MCP Servers and check the status shows Running, confirm the URL was pasted exactly as copied, and confirm the Connector is active in your Vinkius account. A deactivated Connector stops responding.

  • 06

    Can I remove the access later?

    Open Manage MCP Servers and choose Uninstall in the Action Panel. Your Vinkius account keeps the connection and its history either way, ready for the next AI.

  • More questions about ML Experiment Costing? The Connector page answers them. See everything the ML Experiment Costing Connector can do

About the Connector

What ML Experiment Costing adds to your AI.

Once you connect it, your AI gains 4 capabilities: the exact actions it can take in ML Experiment Costing when you ask. The Connector page lists every one of them in detail, alongside the latency we measure in production and prompts worth trying first.

See everything the ML Experiment Costing Connector can do