Review · Software & AI
Promptfoo review
Open source CLI and library for evaluating and red-teaming prompts, models and LLM apps, built to run locally and in CI/CD.
By the DBotMaker research desk · Updated November 7, 2026
- Capabilities (40%)
- 7.2
- Ease of setup (30%)
- 7.6
- Value (30%)
- 9.2
Promptfoo covers the testing half of prompt management rather than the serving half. It is an open source CLI and library: you declare prompts, test cases and providers in a config, run the evaluation, and review outputs in a matrix view in the terminal or web viewer, with automatic scoring against metrics you define. It runs entirely on your machine, talks directly to the model, and slots into CI/CD, so a pull request that changes a prompt can fail if quality drops. Red teaming and vulnerability scanning are built in, with 10k probes a month free in the Community edition. Promptfoo works with OpenAI, Anthropic, Azure, Google, Hugging Face and custom APIs, from any language. What it does not do is store or serve prompts at runtime, so it will not let a product manager change production wording without a deploy. Team sharing, monitoring and SSO need an Enterprise quote.
Who it is for
Engineering teams that keep prompts in Git and want automated tests that block bad prompt changes before merge.
Price
Community edition free and open source; Enterprise custom.
Our open-source tool, perfect for individual developers and small teams.
Quoted from the vendor pricing page on November 7, 2026.
Key features
- Declarative test cases for prompts without writing code
- Matrix views comparing outputs across many prompts
- Automatic scoring with custom metrics
- Red teaming and vulnerability scanning
- CLI, library or CI/CD use with caching and concurrency
Strengths and weaknesses
- Strength: Community edition includes all evaluation features for free
- Strength: Runs locally, so evals talk directly to the model
- Strength: Language agnostic and works with any LLM API
- Weakness: No prompt registry or runtime prompt serving
- Weakness: Team sharing and collaboration are Enterprise features
- Weakness: No published price for Enterprise
Key facts
- Community: Free, all LLM evaluation features, 10k red-team probes a month
- Enterprise: Custom pricing, managed cloud
- On-premise: Custom pricing, complete data isolation
- Usage: CLI, library or CI/CD
Where it ranks
Promptfoo is No. 10 in our Best AI Prompt Management Tools verdict.
Alternatives we scored
- Langfuse9.1
- PromptLayer8.8
- LangSmith8.7
- Braintrust8.5
Visit promptfoo.dev · Questions or corrections: editorial@dbotmaker.io