smigolsmigol/llmkit

Overview

LLMKit is an API gateway for AI agents that tracks and controls spending on language model calls. It enforces budgets, logs sessions, and provides MCP (Model Context Protocol) tools to integrate cost management into agent workflows. Built in TypeScript and available as an open-source project.

Best for

Best for
Developers building AI agents that need to monitor and constrain LLM API spending

Use cases

• Set per-agent or per-user spending limits on LLM API calls
• Monitor real-time costs across multiple agent sessions
• Integrate budget enforcement with MCP-compatible tools

Notes

LLMKit is an API gateway for AI agents that tracks and controls spending on language model calls. It enforces budgets, logs sessions, and provides MCP (Model Context Protocol) tools to integrate cost management into agent workflows. Built in TypeScript and available as an open-source project.

13 stars on GitHub. Last updated 2026-05-28. Licensed MIT.

Use cases

• Set per-agent or per-user spending limits on LLM API calls
• Monitor real-time costs across multiple agent sessions
• Integrate budget enforcement with MCP-compatible tools

Pros

• Clear cost transparency with per-session tracking
• Hard budget enforcement prevents runaway spending
• Open source and self-hostable for full control

Cons

• Small project with 13 stars, limited community support
• Early stage, documentation and examples may be sparse
• Requires self-hosting and manual setup

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.