Migrating from Promptfoo

1. Point eval:local at your existing config

npx @evalguard/cli eval:local promptfooconfig.yaml

EvalGuard reads Promptfoo YAML configs directly. Standard assert types are mapped to EvalGuard scorers automatically — e.g. llm-rubric → llm-grader, model-graded-factuality → factuality, is-json → json-valid, similar → semantic-similarity.

eval:local calls the model on your machine, so set a provider key first — e.g. export OPENAI_API_KEY=… (or whichever provider your config targets). To smoke-test the wiring without a key, add --provider echo, which echoes each prompt back so scorers run offline.

2. Or convert to an explicit evalguard.config.json

npx @evalguard/cli import:promptfoo promptfooconfig.yaml

Writes a fully-converted evalguard.config.json you can review and commit, and prints a summary of any assertion types that need manual attention. The converted output looks like:

// evalguard.config.json (produced by import:promptfoo)
{
  "name": "Imported from Promptfoo",
  "providers": [{ "provider": "openai", "model": "gpt-4o-mini" }],
  "prompts": ["Answer: {{input}}"],
  "cases": [
    { "vars": { "input": "What is 2+2?" },
      "scorers": [{ "scorer": "contains", "value": "4" }] },
    { "vars": { "input": "Capital of France?" },
      "scorers": [{ "scorer": "icontains", "value": "Paris" }] }
  ]
}