eval-framework

Eval framework. Define correct, test against it, get results.

GEDD turns domain-expert judgment into one failure codebook, then derives your evaluation, your training reward and your response gate from it.

A web-based interactive demo for the GuessArena evaluation framework

One-stop CLI for running structured skill evals across all agent harnesses

4-model parallel planning workflow with eval framework — Claude, Gemini, Codex, GLM-5 · OpenClaw ecosystem

Curated AI agent evaluation skills from Microsoft's Eval Guide — plan, generate, run, and interpret eval suites for Copilot Studio agents

Evaluate and trace tool-calling LLM agents, with one guarantee: a tool failure can never look like an empty result.

Evaluation framework for testing LLM outputs locally. Define prompt templates and custom scorers, run evals against OpenAI or Anthropic models, store results in SQLite, and browse via Rich CLI dashboard or FastAPI UI. Lightweight, self-contained, extensible Python library.

Evaluate RAG pipelines end-to-end. Auto-generates Q&A benchmarks from your documents, runs them through your retrieval system, and scores answers on faithfulness, relevance, and context recall using LLM-as-judge. Local vector store, pluggable LLM backend.

Open-source evaluation framework for MCP servers powered by Claude. Auto-discovers tools, generates test scenarios, runs LLM-as-judge scoring to assess correctness and safety, audits for security issues, and visualizes traces in a web dashboard.

Python tool for extracting validated, schema-defined JSON from documents (PDF, HTML, Markdown, plaintext) using LLMs, with source grounding and a built-in eval harness. Works with Anthropic Claude or OpenAI.

CLI tool for evaluating and comparing LLM prompts. Define test cases in YAML, run against OpenAI and Anthropic models, score outputs, and generate HTML reports. A lightweight, local-first alternative to cloud eval platforms.

Open-source evaluation framework for LLM agents. Run head-to-head A/B tests, score with LLM-as-judge rubrics, and visualize results in a Streamlit dashboard. Model-agnostic, self-hosted, zero external infrastructure.

Self-hosted evaluation framework for tool-calling LLM agents. Define tasks in YAML with expected tool calls and scoring rubrics, run against Claude or GPT-4, get scored reports via CLI and FastAPI dashboard. Lightweight, local-first, no lock-in.

Observability layer for multi-step AI pipelines — traces execution, auto-diagnoses root causes, and builds a growing eval dataset from human feedback

Self-hosted evaluation harness for AI agents using rubric-based LLM-as-judge scoring plus deterministic checks. Web UI for test case management, rubric definition, trace inspection, and side-by-side agent comparison.

CLI tool for defining and running LLM evaluation suites against any OpenAI-compatible endpoint. YAML test cases, multi-metric scoring (exact match, regex, LLM judge), and rich terminal reports. Built for fast feedback loops in CI/CD.

Open-source evaluation framework for AI agents. Define test suites with rubrics, run your agent, get LLM-as-judge scores against criteria, inspect full execution traces, and diff runs to catch behavioral regressions.

Define YAML rubrics, run agents through test scenarios, get LLM-judged per-criterion scores with full trajectory traces, and analyze results in an interactive web dashboard.

A FastAPI WebSocket service and CI-friendly CLI for running LangSmith evaluations on demand — bundles LLM-as-judge (Claude) and heuristic evaluators, syncs datasets idempotently, and gates deploys via threshold checks.