Promptfoo

Promptfoo

Promptfoo is an open source command line tool and library for evaluating and red teaming large language model applications, built for developers who want repeatable tests for prompts, models, and retrieval augmented generation instead of manual spot checks. You define test cases that represent your core use cases and failure modes, configure the prompts, test cases, and providers you want to compare, run the evaluation, and then review the outputs in a matrix view or web viewer that places results from many prompts side by side so the best prompt and model choice is visible at a glance. Evaluations in Promptfoo score outputs automatically against metrics you define, and the tool speeds the loop up with caching, concurrency, and live reloading. It works as a command line tool, as a library, or inside continuous integration and delivery pipelines, and it is language agnostic, so teams can use it from Python, JavaScript, or other languages. It connects to OpenAI, Anthropic, Azure, Google, Hugging Face, open source models such as Llama, and custom API providers, which makes it practical to benchmark the same prompts across providers without rebuilding the test harness for each one. Test cases are simple and declarative, so evals can be defined without heavy notebooks or a lot of code, and built in sharing helps teammates review the same results. The second half of the product is automated red teaming and penetration testing for AI applications. Promptfoo simulates real users to uncover application specific vulnerabilities, generating custom attacks for your target that cover direct and indirect prompt injection, jailbreaks tailored to your guardrails, data and personally identifiable information leaks, business rule violations, insecure tool use in agents, and toxic content generation. It integrates with CI and CD pipelines, GitHub, GitLab, Jenkins, model context protocol and agent frameworks, and can run on premises or in the cloud. Findings come back with remediation guidance inside pull requests and developer workflows, so teams can track fixes, close the loop, and keep monitoring continuously rather than treating security testing as a one time audit. Promptfoo runs locally, with evaluations executing on your own machine and calling the providers you configure, and the code is open source so you can inspect and modify it on GitHub. The vendor says the tool was originally built for LLM applications serving millions of users, and its homepage reports adoption by a large share of the Fortune 500 and a community of hundreds of thousands of developers, with no vendor lock in. For prompt work specifically, that combination matters: the same declarative test set that grades prompt quality also becomes the regression suite that catches a prompt change, a model swap, or a new attack pattern before it reaches production.

Similar listings in category

PromptBase

PromptBase

PromptBase is a marketplace for buying and selling AI prompts, built for people who would rather start from a proven prompt than a blank page. The homepage gives access to 330,000 AI prompts for ChatGPT, Gemini, Claude, Midjourney and more, and the site reports a 4.9 rating across more than 44,000 reviews, with more than 520,000 users trusting the platform. Buyers can browse the marketplace by category, with popular sections including Video, Apps and Games, Print and Sell, Photography, Art and Illustrations, Logos and Icons, Marketing and Business, and Holidays. Listings are ranked by sales, views, favorites, uses, and reviews, so the prompts that actually work rise to the top. A free AI prompts section lets you try the format before spending anything. For regular users, PromptBase Select opens access to more than 275,000 Select prompts, priced at 14 dollars for the first month and then 19 dollars a month, with any 10 prompts yours to pick each month. Beyond the marketplace, PromptBase includes a Create area with an AI app builder that runs on credits, an app store, creator communities, and a Hire an AI expert route for custom work. Sellers keep all rights to their prompts, submissions are approved within 15 minutes to 36 hours, and the marketplace takes a 20 percent fee on sales, dropping to 0 percent on sales made through a seller referral link and 10 percent on custom jobs. Payouts run through Stripe or Zoneless on a daily, weekly, or monthly basis. Refunds are available only where a prompt does not work as described, and requests must be made within 24 hours, which keeps quality pressure on sellers.

FlowGPT

FlowGPT

FlowGPT describes itself as the open platform for AI applications, and as a store for ChatGPT characters and prompts. At its core it is a community driven platform for sharing and discovering AI prompts: anyone can browse a large and growing library of prompts, use the ones that fit, and create and share their own for others to build on. The homepage organizes discovery around everyday jobs and interests, with categories that include Job Hunting, Health and Lifestyle, Academic, Business, Programming, Marketing, Image Generator, Prompt Generator, Anime, Game, Productivity, and AI Tools, alongside a wide range of character styles for roleplay and creative writing. That mix is the point of the platform: practical work prompts sit next to entertainment and character prompts, all contributed by the community rather than a single editorial team. FlowGPT documentation explains that getting started is simple: create an account and log in, then explore the library or publish your own prompts. The interface is designed to be user friendly, so finding a prompt for a resume, a study task, a marketing draft, or a coding problem takes a search rather than a long scroll. For prompt creators, FlowGPT offers an audience: share a prompt that solves a real problem and the community can discover, use, and respond to it. For everyone else, it works as a shortcut library, a place to see how experienced users phrase requests for ChatGPT style models before writing your own.