Promptfoo alternative

Independent. Open source. Built for agent code.

Promptfoo was acquired by OpenAI in March 2026 (~$23M pre-acq valuation). Promptfoo remains accessible but is now part of OpenAI Frontier evals. Independence is over — product direction flows through OpenAI's eval-and-test stack.

You're looking for an alternative because (a) OpenAI ownership is a conflict of interest if you ship multi-provider agents, or (b) you need static analysis pre-deploy rather than runtime eval testing.

What Inkog provides

  • Static analysis of agent code — finds bugs BEFORE you write evals against them
  • Multi-provider neutral: LangChain, LangGraph, CrewAI, AutoGen, Pydantic AI, smolagents, Google ADK, Azure AI Foundry, n8n — Anthropic + OpenAI + Google + Cohere + Mistral + open-source backends all supported
  • AGENTS.md governance verification + EU AI Act mapping (not in Promptfoo's scope)
  • Open source (Apache 2.0), independent — no model-vendor strategic alignment

What Inkog doesn't replace

  • Promptfoo's eval-and-test loop for prompt regression testing (running prompts against outputs, scoring them).
  • Red-teaming via adversarial prompt generation — that's a different workflow (we're static; that's dynamic).
  • Most mature teams keep their Promptfoo eval suite AND add Inkog for the static-code layer. The two are complementary.

Try the Promptfoo-displaced workflow in 60 seconds

Paste a GitHub URL. We'll scan it with Inkog\'s 15-adapter IR engine and return findings mapped to OWASP + NIST + EU AI Act.