Today's drop: 2 new tools, 2 listing updates
The directory now lists 133 hand-checked tools. New today: PyRIT, Microsoft's free, open-source toolkit for red teaming AI apps, and Factory Droid, a model-agnostic coding agent whose 1.0 release has the model write small programs to run many tool calls at once. Updated: ChatGPT (switch between two accounts on your phone) and OpenAI Codex (GPT-6.1 Terra, a cheaper fast model). AI for IT & Cybersecurity grows to 48 tools. Pricing only where the official page states it.
Previous drop: Nov 1, 2026 →
Get the weekly dropThe best new AI tools, one email a week. Free.
New in the directory 2
Browse all 133Tool
Category
Price
Status
Link
PyRIT (Microsoft)Microsoft's open-source AI red teaming toolkit (the Python Risk Identification Tool) for testing generative AI apps and models: automated scans, adaptive multi-turn attacks such as jailbreak and prompt-injection attempts, scoring of what got through, and CoPyRIT, a web UI. Built by Microsoft's AI Red Team. About 4,700 GitHub stars and active daily; v1.0 shipped Jul 2026 and v1.1.0 on Sep 4, 2026. Only test AI systems you own or are authorized to test.
Price Open source (MIT); model usage billed by your AI provider
Factory DroidFactory's model-agnostic coding agent, in a desktop app for macOS and Windows, a terminal UI and a TypeScript SDK (@factory/droid-sdk), with cloud and local background agents. Droid 1.0 (Oct 29, 2026) makes Script mode the default on supported models: the model writes a small JavaScript program, run in a sandbox, that makes many tool calls in one request, so raw files and logs stay out of its context. Factory says this cut Droid's cost 15% on Claude Opus 5.5 and that Droid beat other harnesses on its own benchmark runs; those are Factory's numbers.
Price Pro $20/mo; Plus $100/mo; Max $200/mo; Teams $60/mo plus $40/seat
Why these two
- PyRIT (Python Risk Identification Tool) is the open-source, MIT-licensed toolkit Microsoft's AI Red Team uses to probe generative AI systems. It runs automated scans and adaptive multi-turn attacks, such as jailbreak and prompt-injection attempts, against a model or AI app, scores which ones got through, and includes CoPyRIT, a web UI. It has about 4,700 GitHub stars, gets commits almost every day, hit v1.0 in July and shipped v1.1.0 on Sep 4. We already list NVIDIA's garak and promptfoo; PyRIT is the one to learn if your work runs on Azure and Microsoft AI. Only point it at systems you own or are authorized to test.
- Factory Droid is a coding agent from Factory that works with whichever model you pick, in a desktop app for macOS and Windows, a terminal UI and an SDK. Droid 1.0 (Oct 29) turns on Script mode by default for supported models: instead of calling tools one at a time, the model writes a short JavaScript program, run in a sandbox, that makes many tool calls in a single request, so raw files and logs stay out of its context. Factory says that cut cost 15% on Claude Opus 5.5 and that Droid beat other harnesses in its own benchmark runs; those are Factory's numbers. Plans start at $20/mo.
Listing updates
Checked on Nov 2, 2026 against official pages and announcements.
- ChatGPT: account switching is rolling out in the iOS and Android apps (release notes, Oct 30). You can stay signed in to up to two accounts and switch in Settings > Switch account. Chats, memory, files and subscriptions stay separate, which helps if you keep a work and a personal account. It also works on the web.
- OpenAI Codex: GPT-6.1 Terra, OpenAI's fast, cheaper model for computer use and writing, is available in Codex and ChatGPT Work for Plus, Pro, Enterprise and Edu users (released Oct 26). OpenAI says your plan's allowance covers more tasks with Terra than with GPT-6.1 Sol or GPT-6 Astra. API price: $1 per million input tokens and $5 per million output.
Category changes
- AI for IT & Cybersecurity grows from 47 to 48 tools with PyRIT. Coding & Vibe Coding adds Factory Droid.
Checked, not listed
- DeepSeek dsh-superagent (Oct 31, MIT): an agent that works on one GitHub repo and answers requests in issue threads, often with a pull request. DeepSeek calls it a research prototype, not an official product, so we're waiting.
- TokenEconomizer (Microsoft Research, Apache-2.0): a plugin meant to cut coding-agent token use in repos you use often. Too new to show real use yet; the savings numbers are the authors' own.
- Vercel agent-desktop: an open-source CLI that lets agents control desktop apps through accessibility trees. Interesting, but early.