LIVE
Google Introduces "Gemini 4 Carbon" as Next Step in Its Model Lineup28/10/26 · Google|DeepMind unveils Genie 4, a new generation of world model27/10/26 · Google DeepMind|Anthropic files IPO prospectus, with risk disclosures taking center stage26/10/26 · Anthropic|Robot policy trained only in simulation matches top real-world systems26/10/26|OpenAI unveils GPT-6.1 Terra, a model built for computer use26/10/26 · OpenAI|Google's AI gives evasive answers about who won the 2020 US election28/10/26 · Google|NVIDIA introduces Nemotron 3.5 Transcribe, a promptable multilingual transcription model28/10/26 · NVIDIA|OpenAI and ServiceNow deepen partnership on enterprise workflows28/10/26 · OpenAI|Failed startups' data finds a second life training AI models28/10/26|Anthropic expands AI for Science program to 20 foundations and research institutions28/10/26 · Anthropic|Anthropic shares early results on using Claude to accelerate drug development27/10/26 · Anthropic|GEPA-Curator: Filtering Failures to Better Train LLM Agents27/10/26|Google Introduces "Gemini 4 Carbon" as Next Step in Its Model Lineup28/10/26 · Google|DeepMind unveils Genie 4, a new generation of world model27/10/26 · Google DeepMind|Anthropic files IPO prospectus, with risk disclosures taking center stage26/10/26 · Anthropic|Robot policy trained only in simulation matches top real-world systems26/10/26|OpenAI unveils GPT-6.1 Terra, a model built for computer use26/10/26 · OpenAI|Google's AI gives evasive answers about who won the 2020 US election28/10/26 · Google|NVIDIA introduces Nemotron 3.5 Transcribe, a promptable multilingual transcription model28/10/26 · NVIDIA|OpenAI and ServiceNow deepen partnership on enterprise workflows28/10/26 · OpenAI|Failed startups' data finds a second life training AI models28/10/26|Anthropic expands AI for Science program to 20 foundations and research institutions28/10/26 · Anthropic|Anthropic shares early results on using Claude to accelerate drug development27/10/26 · Anthropic|GEPA-Curator: Filtering Failures to Better Train LLM Agents27/10/26|
AppRisingofficial publisher

Promptfoo

App : A product you use directly

Prompt, agent, and RAG testing and red-teaming, CI-integrable.

The testing tool to adopt before putting an AI into production; acquired by OpenAI, still MIT licensed.

Official site GitHub · 26,392 ★
Overall score
9.0 /10
Security: nothing to flag
Utility
8.5
Reliability
10.0
Security
No caveat on record
Ease
5 min · Technical

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

Deploying an AI application without prior testing exposes it to real risks: prompt injections, personal data leaks, guardrail bypasses, and unexpected agent behaviors. Promptfoo addresses this by automating security and evaluation testing before deployment.

In practice, the tool generates thousands of context-aware attacks tailored to your application (direct and indirect injections, jailbreaks, business rule violations, insecure tool use in agents), compares responses from different models side by side, and produces vulnerability reports. It integrates into CI/CD pipelines (GitHub, GitLab, Jenkins) and can scan pull requests to detect AI-related security issues.

Promptfoo is aimed at developers and security teams building applications on top of LLMs, agents, or RAG systems. It is used by 156 Fortune 500 companies across sectors such as healthcare, telecommunications, and retail. Evaluations run entirely locally, meaning prompts never leave your machine.

However, the tool requires Node.js 22.22.0 or higher (Node.js 24 LTS recommended) and a degree of familiarity with the command line and YAML configuration files. It is not designed for non-technical end users.

What it offers

  • ·Automated prompt and model evaluation with side-by-side comparison across providers (OpenAI, Anthropic, Azure, Bedrock, Ollama, and more).
  • ·Automated red teaming that generates custom attacks to detect over 50 vulnerability types, including prompt injections and jailbreaks.
  • ·Security testing for agents and RAG systems, covering data leaks, business rule violations, and insecure tool use.
  • ·Integration into CI/CD pipelines (GitHub, GitLab, Jenkins) for continuous checks at every deployment.
  • ·Code and pull request scanning to detect AI-related security and compliance issues.
  • ·Vulnerability reports with remediation steps delivered directly into developer workflows.
  • ·100% local execution of evaluations: prompts and data never pass through external servers.
  • ·Real-time threat intelligence powered by a community of over 300,000 users.

In practice

Price
Freemium
Commercial use
Commercial use: yes
Available on
cli, docker
Licence
MIT

Security and data

Rating
Security: nothing to flag
Your data
Data stays local
Known incidents
Rachat par OpenAI en mars 2026.
Maintenance
Active
Last release
25 October 2026

Links and repository checked on 25 October 2026 · added on 1 May 2023

How to install Promptfoo

1. Verify that Node.js 22.22.0 or higher is installed on your machine (Node.js 24 LTS recommended). 2. Install Promptfoo globally via npm: npm install -g promptfoo. 3. Alternatively, install it via Homebrew (brew install promptfoo) or pip (pip install promptfoo), or use npx promptfoo@latest without installing. 4. Set your API key as an environment variable, for example: export OPENAI_API_KEY=sk-abc123; never write it in plain text in a versioned file. 5. Initialize an example project with promptfoo init --example getting-started, then run promptfoo eval and promptfoo view to see the results.
npm install -g promptfoo

In the same category

Whole category Frame (methods, guardrails) →