Connect Your AI Agent to Cloptima
Bring Cloptima's LLM spend and AI gateway governance into the AI tools your team already uses — connect directly to our hosted MCP endpoint, or run it locally with the Cloptima CLI. Same tools, either way.
What You Can Do With It
Investigate AI Spend and Governance
Bring LLM spend and AI gateway governance directly into your agent workflow.
What's our LLM spend by model this month?
Which prompts triggered a guardrail block?
Show me open policy violations on the AI gateway.
Operate the Gateway
Issue and rotate virtual keys, check which policy governs an app, and review alerts. Every change asks for confirmation first.
Create a virtual key for the support-copilot app.
Show the effective policy for the support team.
Which LLM alerts are still open?
Connect However You Work
Same tool set, two ways to connect — pick hosted for zero infrastructure, or local for offline and air-gapped use.
Point any MCP client at our hosted endpoint with a token.
Or run the server locally for full control over where it lives.
Use read-only mode for safer agent access either way.
What Your AI Tool Can Access
AI Gateway Spend and Governance
Query LLM spend by model, provider, and app, inspect guardrail and policy violations, and review approvals — the largest single category of MCP tools.
Attribution and Unit Economics
Check attribution quality, export showback data, and report cost per ticket, workflow, or customer.
AI Spend Findings
Let agents review ranked AI spend findings, such as anomalies, unattributed spend, and retry waste, and update their status.
Keys, Edges, and Pricing
Issue and rotate virtual keys, register edge gateways, and manage contract price sheets. Every change needs explicit confirmation, with an optional dry-run preview.
Alerts and Caching
Review LLM alert events and rules, preview exact-cache hit rates, and approve semantic cache classes.
Works with MCP Clients
Claude, Cursor, Cline, and custom agent runtimes can all connect — over our hosted endpoint or local stdio transport.
50+ AI Tools, One Contract
LLM spend, attribution, governance, pricing, alerts, and gateway keys — the same tools whichever way you connect.
Token-Based Auth
Reuse your existing Personal Access Token, with the same expiry, scoping, and revocation controls from your Cloptima settings.
Hosted or Local, Your Choice
Connect directly to our hosted endpoint for zero infrastructure, or run cloptima mcp serve locally for offline use. Use --read-only for a safer default either way.
Quick Setup
Hosted Endpoint
Endpoint: https://api.cloptima.ai/v1/mcp
Authorization: Bearer <your-personal-access-token>
# Any MCP client that supports Streamable HTTP
# can connect directly — nothing to install.Local Server (CLI)
cloptima mcp serve
cloptima mcp serve --read-only
# Claude Desktop / Cursor / Cline MCP server entry
{
"mcpServers": {
"cloptima": {
"command": "cloptima",
"args": ["mcp", "serve"],
"env": {
"CLOPTIMA_ACCESS_TOKEN": "pat_..."
}
}
}
}Typical prompts include: "What's our LLM spend by model this month?", "Which policies blocked requests this week?", and "Which LLM alerts are still open?"
Questions teams ask
Bring AI Spend Intelligence to Your Agents
Bring LLM FinOps, governed model access, and guardrails into one operating model.