# eval-builder MCP server

> Tests LLM judges for position and padding bias and converts conversation logs into reusable evaluation suites.

- Repository: https://github.com/Abelo9996/eval-builder
- Page: https://hostedmcp.org/servers/abelo9996/eval-builder
- Transport: —
- Stars: 0
- License: MIT

## FAQ

### Can I use my own logs to build an evaluation suite?

Yes, you can ingest logs from OpenAI, Anthropic, Langfuse, OpenTelemetry, or JSONL to build a suite.

### Does it work with local models?

Yes, it supports local models via Ollama and runs for free.

### Where can I export the evaluation suites to?

The tool exports to existing platforms like promptfoo, DeepEval, Inspect AI, or plain JSONL.
