DAILY DROP · OCT 29, 2026

Today's drop: 3 new tools, 2 listing updates

The directory now lists 125 hand-checked tools. New today: BotGauge, which attacks the AI agents you build to find where they break before users do; Project Quarry, Cisco's new open-source rule engine for spotting risky patterns in AI agent sessions; and LFM3, Liquid AI's small open-weight models that run on a laptop or phone. Updated: ChatGPT (locked chats) and Ollama (LFM3 in the library). AI for IT & Cybersecurity grows to 41 tools. Pricing only where the official page states it.

Previous drop: Oct 28, 2026 →

Get the weekly dropThe best new AI tools, one email a week. Free.

New in the directory 3

Browse all 125
Tool
Category
Price
Status
Link
BotGaugeSecurity and reliability testing for AI agents you build. It runs adaptive red-team attacks (prompt injection, policy bypass, tool misuse, multi-turn manipulation), replays each failure with a full trace and turns it into a regression test, then keeps watching the agent in production. Connects through an SDK to OpenAI, Anthropic, LangChain, LangGraph, LlamaIndex, CrewAI and AutoGen. Launched as the #3 product of the day on Product Hunt on Oct 28, 2026.
Price See site
See site
Project Quarry (Cisco)Open-source detection rules for AI agent sessions, from Cisco AI Defense. You write rules that match patterns across messages, tool calls, MCP activity and the order they happen in (for example, a read of a sensitive file followed by an outbound send), and a match returns the evidence with secrets masked. Use it as a runtime guardrail, to review past agent traces or during incident response; it ships a plugin for Codex, Claude Code and MCP hosts plus a starter rule pack. Apache-2.0, released Oct 28, 2026, and still alpha, so the rule syntax may change before v1.0.
Price Free and open source (Apache-2.0)
FREE TIERchecked 10/29
LFM3 (Liquid AI)Liquid AI's family of six small open-weight models (about 2B parameters each) for text, vision with video input, audio, combined voice and vision, and fast yes/no style decisions. The text model fits in about 1 GB at 4-bit, has a 256K-token context and is tuned for local agents; every model can be fine-tuned on a single GPU. Runs with llama.cpp, Ollama, LM Studio, MLX and vLLM. Released Oct 27, 2026; benchmark claims are Liquid AI's own.
Price Free open weights (LFM Open License 1.0); Liquid API also available
FREE TIERchecked 10/29

Why these three

  • BotGauge covers testing for agents you are building. We already list garak and promptfoo for red-teaming models and prompts; BotGauge aims at agents that call tools and touch real data. It runs adaptive attacks (prompt injection, policy bypass, tool misuse, multi-turn manipulation), replays each failure with a full trace, turns it into a regression test and keeps retesting as the agent changes, then monitors it in production. It hooks in through an SDK for OpenAI, Anthropic, LangChain, LangGraph, LlamaIndex, CrewAI and AutoGen. It was the #3 product of the day on Product Hunt on Oct 28 with about 250 upvotes. The launch post says you can try it free; we could not find published pricing for the agent product, so check before you commit.
  • Project Quarry is the detection side of agent security. Tools like HOL Guard block a single risky action; Quarry rules can match a sequence across a whole session, such as a sensitive file read followed by an outbound send, and return the evidence with secrets masked. Cisco AI Defense open-sourced it on Oct 28 (Apache-2.0) with a plugin for Codex, Claude Code and MCP hosts and a starter rule pack, and The Register compared it to Snort and YARA for agents. It is brand new (v0.1.0, a few dozen GitHub stars so far) and labeled alpha, so expect the rule syntax to change before v1.0.
  • LFM3 is a new option for local models next to Ollama's library. Released Oct 27, it is six roughly 2B-parameter open-weight models (text, vision with video input, audio, a combined voice and vision model, and two decision models), all fine-tunable on one GPU. The text model needs about 1 GB at 4-bit and has a 256K-token context, which makes it a realistic choice for a home lab laptop. The text model had about 16,000 Hugging Face downloads in its first days, and launch partners include AMD, NVIDIA, Qualcomm and Vercel. Liquid AI's benchmark tables are its own, not independent tests, and the LFM Open License has its own terms, so read it before commercial use.

Listing updates

Checked on Oct 29, 2026 against official pages and announcements.

  • ChatGPT: since Oct 20 you can lock a conversation from its menu and reopen it from Projects with a four-digit PIN or your device's sign-in. Locked chats drop out of regular history, pinned chats and search. OpenAI notes this hides chats but does not add end-to-end encryption or change data retention, and resetting a forgotten PIN permanently deletes the locked chats.
  • Ollama: Liquid AI's LFM3 models are now in the Ollama library, so you can try a small on-device model with one command.

Category changes

  • AI for IT & Cybersecurity grows from 39 to 41 tools. The Breaking into IT guide adds Project Quarry (write a detection rule for your own agent's traces, the same way SOC analysts write rules for logs) and BotGauge (see how agent red-teaming works).

Checked, not listed

  • Dedica (Product Hunt #1 on Oct 28): checks whether AI agents can use your website. Useful for site owners, but outside this directory's IT and AI-tools focus for now.
  • Zro (Product Hunt #10): private inference with uncensored security models for red-team work. Too new to judge, and uncensored models need more vetting before we point beginners at them.
  • GitHub code scanning for AI applications (public preview Oct 28): flags prompt injection and excessive agency in pull requests, but it needs a GitHub Advanced Security license, so most readers can't try it.
  • Enterprise IT and SOC launches this week (ControlUp Pulse AI Resolve, Virtana AOEM, Swimlane Turbine detections, Wizard Cyber CYBERSHIELD AI): sales-led products with no self-serve tier.