Humanbound
Humanbound continuously tests AI agents and chatbots with adversarial attacks, tags findings to OWASP LLM and Agentic Top 10, and scores security posture.
1
DEV TRENDS / CATEGORY
Agent ecosystem · Agent evaluation & safety
61 new projects first discovered during Oct 21, 2026 – Oct 27, 2026, deduplicated across 6 sources.
596 projects first discovered Aug 5, 2026 – Oct 27, 2026
DTDevTrends classificationOriginalOriginal project or source tag
61 new projects6 sources
Agent Index
Agent Index is a free, sourced directory of about 404 AI agents for compliance, security, risk and business operations, plus about 483 independently verified audit firms (SOC 2, ISO 27001, PCI DSS, HITRUST, CERT-In and more). Compare tools side by side, filter by framework, region and integration, and find auditors by country and city. Not a ranking.
1
lfg
LFG turns Claude Code into an engineering pipeline you direct instead of type into. Describe a change in one sentence and specialist agents plan it, build it test-first, and check it with lint, types, tests, security scans and an expert review. You review by talking to the diff, see what a senior engineer would ask, and decide what ships. It runs on your machine, never pushes or reads your secrets, and is free and open source.
2
Reins —— 自托管零信任远程运维平台
Reins — Take the reins of your fleet. Self-hosted zero-trust remote operations for Windows (first-class) & Linux: one Go binary, zero inbound ports, PowerShell/GUI/browser automation, mTLS+RBAC+audit chain, AI-agent rules gate.
2
NVIDIA / SkillSpector
Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, security risks, prompt injection, data exfiltration, and supply-chain risks in Claude Code, Codex, and MCP skills before you install them.
22.1K
CovaSyn Chemistry Tools
All-in-one chemistry MCP server for pharma and biotech: over 250 tools in 18 families for spectra, stability and toxicology. Try it free in Claude, ChatGPT, Cursor. 14 to 41% without CovaSyn, 76 to 92% with CovaSyn. Measured on MolecularIQ, a benchmark we did not design (arXiv:2601.15279, Klambauer Lab, Institute for Machine Learning, JKU Linz).
1
输入一篇论文，生成带引用的商业化初筛报告
AI multi-agent tool that turns research papers into cited commercialization and tech-transfer assessments: TRL, patent landscape, market and risks. CrewAI + FastAPI, with a reproducible benchmark.
774
KageX
KageX is an offensive AI security company. Our flagship FreakLabs teaches AI red teaming by doing, with 20 hands-on labs against live chatbots, RAG systems and agents, a 9-level course and a verifiable AI Red Teamer certificate. There is a free tier, a free CTF called SentinelGoat, and open-source tools for testing your own AI apps. Built for pentesters, security engineers and developers shipping LLM features.
2
I built a tool to audit Shopify stores for AI agent and Google Shopping
No project description yet.
—
Markdown is 40× more token-efficient with SQLite
Hi HN, I'm Richard, founder of Native. I'm working on open-source context ownership, to stop AI centralisation: https://github.com/withnative/native My first focus has been making Markdown vaults powerful enough so that teams can self-host all of their context (messaging, docs and projects), without it being externally fragmented or rate-limited. I'm sharing some early results: Markdown vaults can be 40× more efficient with SQLite! https://www.withnative.ai/research-and-posts/markdown-contex... Here's some of the main ideas: - SQLite or Markdown folders, you pick. You can project the same authoritative event log into either form; more projection formats will follow (e.g. maybe neo4j; we do use a graph structure). - Empirically: SQLite is better for agents! On a 120 trial evaluation for 11,000 records, agents use 40× fewer tokens with SQLite vs Markdown, per correct answer. - Everything is a file or SQLite row. Messages, documents and projects: all can be modelled in your vault. You get local-first collaboration with teammates, synced to a cloud of your choice. - You can render your vault with arbitrary UI. Your WhatsApp-like UI speaks to your teammate's Slack-like UI; both are just views over the same database. A few things I've learned building this: - Agents do pretty well even without SQL if you give them strongly typed Markdown frontmatter. This was one arm of the trial I ran. I didn't have any clear expectations, but it was cool to watch: agents ended up writing Python scripts to do SQL-like queries. - Cumulatively, cached input tokens can get expensive. I feel silly in hindsight here. I'd gotten used to letting my sessions run long, "It's all cached so it's very cheap". But when I looked at my results here, over 2/3 of the token bill on long sessions was cached input... - "Markdown vaults don't scale as well as databases" isn't as self-evident as I think. This is why I decided to benchmark it... I've had some very blank stares when I try to explain how Markdown files might break at true agent-scale volume. Unfortunately, the dataset is private, because it's two months of my own work. I am considering open-sourcing it even so, but there is definitely some embarrassing stuff in there. TBD... I know that independent result verification matters! The core engine that I evaluated is open-source, with more to be released soon: https://github.com/withnative/native If you're interested in trying connected context on an open engine, you can get updates from our website or email me directly: richard@withnative.ai https://www.withnative.ai/
—
Hashling, an open-source Mac security agent your own AI can talk to
No project description yet.
—
Agent Payment Node：AI 代理的本地支付工具
Open-source local-first payment runtime for AI agents
0
开源了一个 MCP 治理网关： Jev 判 0.94、d3 判 0.70，两个判官说不同的话时怎么办
给 AI agent 做了个 MCP 执法网关（ Rust ），核心问题：当两个决策模型对同一件事给出不同判断时，怎么安全处理？ 实测：同一个越权场景（ coding agent 试图 merge 夹带无关变更的 PR ）： - Jev 给 0.94 置信度 → 准入层 Accept → 授权挂起 → DENY - d3 （ vLLM SR ）给 0.70 → Hold → 什么都不改 → ALLOW 两个都对。分歧不是 bug ，是校准本性（ d3 ECE 4.3 vs Jev 8.0 ）。治理层的价值在于让分歧安全。 网关做的事：每个 tools/call 先过 r2r_decide 门控，ALLOW 转发上游，DENY 返回 isError + 证据文本（相位 + 规则 + 事件 ID ），全程审计到 EventStore 。 零 API key 复现： cargo run -- fixture cargo run -- fixture --provider d3 GitHub: （ Apache-2.0, v0.2.0 ）
Original
—
incursio
AI agents that audit your code for security vulnerabilities, on your Claude or ChatGPT subscription or on local models with Ollama. Based on Strix. Join the open beta.
3
Vibefilo
Security and GDPR checks inside Claude Code, Cursor and Codex. Find leaked secrets and RLS gaps, then verify every fix. Your code stays on your machine.
1
Browser Extractor
Inspect any live website like a design file. Explore styles, edit elements in place, audit UX and accessibility, and turn your changes into actionable context for your coding agent. All from a free Chrome extension.
2
Open_SwarmForce
SWARMFORGE is an agentic CI/CD gate for the change path that matters: it reviews a committed or working-tree diff, can group and fix confirmed findings in isolated sandboxes, re-reviews the result, then requires deterministic repository checks before it publishes a PR. It runs with OpenAI Codex/GPT, any OpenRouter model, local open-source models (Ollama, LM Studio), or Claude Code (claude -p, opt-in) — with cost-aware routing measured on an L0–L6 benchmark ladder.
9
Sage 7
Your AI writes the code. Sage 7 makes sure it builds the right thing: map the whole product, brief Lovable, Cursor or Claude Code with living context, and audit every push against the plan.
1
Vesalius
Vesalius reads your repository once, right on your Mac, and remembers how everything connects: which file uses which, who calls what, where a feature actually lives. Claude Code can then ask it directly and get answers with exact file and line references, usually in one call instead of a chain of searches. In our benchmark that cut token use roughly in half with no measurable change in answer quality. Vesalius itself never uploads your code.
8
OrqLabs
OrqLabs is a no-code platform where 22 specialist AI agents (Claude & OpenAI) create social content, answer customers 24/7 on WhatsApp and Telegram, find and score leads, audit SEO and rewrite your strategy from real GA4 data. Every publish or outreach step waits for your approval, and every AI call is metered against a monthly budget cap you can't exceed. 109 templates, 75 integrations, works inside Claude and ChatGPT via MCP. Free plan, no credit card.
2
外链检测：提交网站之前，先查这个外链来源值不值得发
Check a link source before you submit to it, verify the listing, and audit your own backlinks for link-spam patterns. A CLI and an agent skill.
28
AgentTrust – Zero-trust security and human approvals for AI agents
Enterprise Zero-Trust Security, Cascading Delegation, Human Approvals & Verifiable Audit Infrastructure for Autonomous AI Agents
17
anlyon.dev
A free index of the official MCP registry. Each tool is sorted by what it does: reads, changes things, can't be undone, or deletes or cancels. Paste your MCP config to audit it, pin a lockfile that fails CI when tools change, and follow change feeds.
1
AXXESS TRIaxis
Enterprise AI for the 80–90% of organisations it prices out. AXXESS is a company brain for small businesses, clinics, schools and nonprofits: it remembers everything scattered across WhatsApp, email and docs, and AI agents act on it, with humans approving what matters. Features: company memory, AI agents and an AI phone operator, CRM, tasks, approvals and audit trails. On web, Android and iOS.
2
Papermark Data Rooms
Data Rooms 2.0: launch a secure data room in under 60 seconds. Encrypted, fully certified, with file permissions, built-in Q&A, a full audit log and analytics. White-label it on your domain, or let agents run it via API, CLI and MCP.
93
Silda
SILDA sits between your OT devices and IT network, inspecting every packet inline without agents or hardware appliances. It detects threats per-asset using online ML, maps them to MITRE ATT&CK for ICS, and generates cryptographic audit chains for NIS2 and IEC 62443 compliance, all without cloud dependency or touching a single PLC or controller. Built for manufacturing, energy, water, and oil and gas.
1
SecureMID
SecureMID helps teams assess the security of AI agents and identify potential weaknesses before deployment. Built for AI developers, security researchers, and teams working to build safer AI systems. Explore our platform, evaluate your workflows, and help shape the future of AI agent security.
1
Neurion
Neurion brings AI chat, project coding, browser automation and security tools into one Windows desktop workspace. Choose Neurion Fast or Neurion Pro and work with your files, custom agents and connected tools. Account and paid subscription required.
1
Argus- open source security platform to protect against agentic attacks
Open-source agentic security operations platform:
83
AgentSkill – Yara-L-Skill
An Agent Skill that teaches AI coding agents to write, review, tune, and convert YARA-L 2.0 detection rules for Google Security Operations (SecOps, formerly Chronicle).
—
HyperLevel
HyperLevel replaces the tools solopreneurs and agencies rent for hundreds a month: CRM, funnels, inbox, automations, courses, helpdesk and invoices – in one app. Self-host free (AGPL), unlimited users, 0 € per seat. Via MCP (340+ tools) or the REST API, your agents and scripts run it – with the same permissions, audit log and undo as your team. Every module is open and built to be modified: fork it, extend it, send a PR – contributions are very welcome.
2
Borg
Borg is your personal assistant, employee, developer, analyst, and sidekick. Security-first by default.
3
VibeDoctor.io
Apps built with Claude Code, Cursor or Codex run and look finished, but nobody has read what the agent wrote. Paste one line into your agent and it checks its own code as it writes. Every scan covers security (secrets, auth, packages) and code health, with one ranked report in under 2 minutes. Vibe X-Ray sends only the files a change touches, so prompts use fewer tokens. Every push is checked, and your live site is watched for speed on a phone and uptime around the clock.
3
Comity
A vendor-neutral coordination layer for teams' AI coding agents: shared awareness, conflict detection before code is written, policy enforcement through agent hooks, and an audit trail.
1
Vitals：按应用而不是进程看Mac的CPU、内存和耗电
自荐一个Mac工具：Vitals，活动监视器的替代品。 活动监视器里有上千个进程，Vitals把它们归到所属的应用下，大约60行：Chrome的几十个辅助进程算作一行，每个应用一个CPU、内存、能耗、磁盘和网络数字。 保留30天的历史，可以看昨天下午是什么拖慢了Mac 某个应用占用过高时提醒 按项目文件夹列出忘了关的开发服务器、端口和Docker容器 自带MCP服务器，Claude、Cursor可以直接问Mac现在的状态 简体中文、繁体中文界面 3天免费试用，不用注册、不用绑卡，之后一次买断。 少数派的评测：
Original
—
Niyoos
Niyoos sits between your AI agents and LLM providers (OpenAI, Anthropic, Gemini) as a proxy, so it can enforce — not just watch: a tamper-proof audit log, real-time cost tracking per agent, an instant kill switch, and drift alerts when behavior changes. Works with n8n, Zapier, Make, or custom code — point your agent at Niyoos, nothing else changes.
1
Distributed, deterministically secure agentic swarm runtime. Enclawed
A Deterministic Security Runtime for LLM Agents on MPI and Batch-Scheduled HPC Systems: Consensus-Governed, Hardware-Rooted Supervision of Every Rank
—
Benten
Benten is an audio evaluation and analytics platform built specifically for Voice AI agents. It connects directly to voice AI platforms to discover your agents, process call audio, and measure conversation quality, non-intrusive speech quality (NISQA MOS), response latency, and speech dynamics.
1
Secure your AI agents on Mac
Plango is a Mac native security suite for anyone running AI coding tools. It works by running a custom local security model that evaluates threats against the lethal trifecta (data access, communication, exposure). Some cool features: - Restrict Agent access by type of data (ssh keys, env files) or finder tags (finally a good reason to use the color tags in finder). - On device security model, but can switch to another local agent or bring your own key. - Review your Mac's privacy and access, apps, startup & extensions. - Automatically move env secrets into keychain. - Network blocks/allow lists, see what apps and agents are accessing. Stop them from data exfil. - Check skills and plugins for bad behavior. - Create custom rules for all of these things. - Log every action, request, block etc in an easy dashboard. - Vuln scan of all projects on your Mac. About us: I'm Fred, long term hacker first time shower. My co-founder Rich and I care about privacy, security and believe that AI software development is the future. We have two decades of combined experience in cyber, NLP and AI and are working on this tool full time. Given how fast cyber is evolving we're pushing changes to this on a daily basis. Please download and trial for free! We're here to answer any questions about the app or cyber/AI.
—
Plango
Secure your Mac while building with AI. Plango monitors AI coding agents, protects sensitive files and secrets, blocks suspicious network activity, scans projects for vulnerabilities, and gives you control over what your agents can access. Powered by an on-device AI security model. Private by default.
82
GroundTruth OSINT
GroundTruth OSINT is a free live dashboard that watches 180+ public feeds (international newswires, official releases, specialist security publications) and clusters the day's reporting into the most significant global events on a 3D globe. Every event shows left, center, and right coverage side by side, with a triage queue, per-event market impact, satellite imagery, and a free daily morning brief. Free public API and MCP access, no key required. No account, no paywall, no tiers.
1
Watchdog — Guardian Agent Supervision
Microsoft (Oct 19, 2026) announced Guardian Agents that "watch and moderate every action of every agent, before it happens, 24/7" — but published no implementation. Watchdog is the drill kit: a FastAPI guardian that scores every agent action and routes it to ALLOW, STEP-UP, HOLD-FOR-HUMAN, or HALT-AND-ISOLATE. Policy template, runbook, identity registry, risk rubric, containment playbook included. 17 tests passing. Not cost routing, not an audit — runtime supervision by safety.
1
Brand Identity API
Give it a domain and it hands back the company's brand: logos for light and dark (SVG first), HEX colors by role, and the fonts the site really loads. Go deeper for 21 sections: the design as a DESIGN.md, the voice as a brand.md, plus pricing, products, competitors, open roles, contacts and investors. Even its email security. Anything we read off a page comes with the exact quote and its URL. No quote, no fact. Call it over REST or MCP. 50 lookups free, then $39 for 10k a month.
3
SlopSquat：面向 AI 辅助编程时代的依赖包幻觉投毒（Slopsquatting）轻量级开源嗅探工具
面向 AI 辅助编程时代的依赖包幻觉投毒（Slopsquatting）轻量级开源嗅探工具。
Original0
AI Agent Security Audit Console – offline audit in 15 min
No project description yet.
—
InterOdin
Odin v1 is an AI model API built for security work, where general-purpose models often refuse legitimate tasks like exploit analysis for detections, YARA/Sigma drafting and CTF payloads. It works with OpenAI Chat Completions, OpenAI Responses and Anthropic Messages, so it plugs into Claude Code, Cursor and Codex with just a base URL and key. Prompts and completions are not stored. Prepaid credits, no subscription, no expiry.
1
BWorlds
You built an app with Lovable, Bolt, Replit, v0 or Claude Code, and now people depend on it. BWorlds reviews it the way a senior engineer would: security, reliability, data, performance, costs, code quality. Then it watches the app 24/7 and sends you one email each morning with what happened overnight and a fix you paste into the tool you build with. Once you publish, it checks the fix landed. The free plan covers the full audit and monitoring on as many apps as you like. No credit card.
5
Embodsi – Embodied Super Intelligence
embodsi is a generative AI MMO experience that runs in the browser. It allows people to communicate (text, audio, video, avatar emotes and movement) and collaboratively build 3D spaces and games via LLMs connected to harnesses and sandboxes for object modeling, space layout, game logic, and driving persistent avatars with shared context and memory. It’s different from existing 3D social apps (AFAIK) because it enables people to build multiplayer virtual worlds, games, and characters with generative AI right from their browser (desktop, mobile, or VR), in nearly infinitely-extendable recursively tiled interconnected spaces, while allowing people to interact in a public commons or easily create a private channel to enjoy the same spaces with their chosen friends and community. The brief backstory is I was running my personal “pelican benchmark” [1] on GPT-6 Astra, which when I ran on Gemini 3 Pro last November couldn’t even put heads on 3D models the right side up [2], and was astounded by the initial results. I gained inspiration for an art project and social experiment that was going to have nine interconnected, benevolent virtual worlds that people could explore and hang out in with AI characters, but by the time Opus 5.5 was released I was gaining so much momentum and having so many ideas and so much fun building I just kept building more and more with the constraint of minimizing MVP scope lifted by agentic coding. I let the project become an ever-evolving wishlist of capabilities for a more interesting social environment as I became increasingly fascinated with the prospect of embodying creative machine intelligence in 3D space with people, realizing that rather than just limiting to building the spaces and characters I imagined I might enable people to co-create endless spaces and characters in a shared world. There’s so much to do still but I’m sharing now even if it feels a little too soon (this is the first public announcement of embodsi’s existence) because I think it’s better to build in the open so people might guide the development and help it evolve in a good direction as model capability improves at a jaw-dropping pace. For people that have the budget to build I hope to see you in embodsi creating amazing things! And for everyone regardless of means (I’ve tried to keep as much free and accessible as financially feasible) I hope you meet amazing people, discover awesome spaces, and enjoy your time in embodsi :) I’ll be around here and embodsi to answer any questions! Thanks for your feedback and critique. - Laser P.S. SI != AI — but an emergent property of many people and what we create cascading and recursively interacting [3] [1] https://news.ycombinator.com/item?id=50074156 [2] https://x.com/lasernite/status/1991021951301923107 [3] https://en.wikipedia.org/wiki/Gaia_hypothesis
—
OhShit!
OhShit! turns public security advisories and breach reports into a feed developers can act on. Each incident has a severity level, affected products, source links and a cleanup checklist. Check supported runtime versions, or follow incidents through RSS, a free JSON API and an MCP server. The code and incident data are open source.
1
Vulnify
Vulnify decides each sensitive AI agent action before it runs: ALLOW, REVIEW or BLOCK, with a 0-100 score, the reasons and a hash-chained audit log. Node and Python SDKs, monitor mode, fail closed by default. The Developer plan is free.
2
AI Shop Audit：免费检测 ChatGPT、Perplexity 等 AI 助手能不能从你的 Shopify 店里买东西
是什么：现在 ChatGPT、Perplexity、Google 都在给用户推荐商品，今年 Shopify 还向浏览器里的 AI Agent 开放了结账。但 AI 能不能真的读懂你的商品、找到购买按钮，取决于店铺页面本身。这个网站就是按 AI 的视角去扫你的店。 免费检测（不用注册、不留邮箱，约 3 秒出结果）： robots.txt 有没有把 ChatGPT、Perplexity 的爬虫挡在外面 有没有给 AI 看的 llms.txt 商品页有没有机器可读的价格和库存数据 需要的话也有 $29 的完整版：扫最多 8 个页面，给评分和按优先级排的修复清单，附可直接粘贴的 Shopify 代码，约 10 分钟发到邮箱。不用装插件，也不需要后台权限，只看公开页面。 适合：做独立站 / Shopify 出海的卖家。目前只支持 Shopify，网站为英文。
Original—
DoneMark
DoneMark provides human accountability for AI-generated code. Run your favorite coding agents (Cursor, Claude Code, Devin), while DoneMark generates cryptographic audit trails and 4 automated proofs per line to let you sign off on intent without breaking production.
1
AgentTrust OS
AgentTrust OS is the runtime trust layer for AI agents. It sits between an agent's decision and the business action that follows: certify agents before they ship, check every execution in real time (approve, retry, escalate or block in under 20 ms, no LLM needed), and keep a tamper-evident audit trail. Wrap an existing agent with one decorator. Data can stay inside your network. Built for developers shipping agents and for security and compliance teams who must prove those agents behaved.
1
Fulcrum Lovable Audit
Built your app with Lovable? Add its URL, connect the GitHub or drop the ZIP export, and get prioritized findings on security, scalability, running costs, code quality and visibility, each with a fix prompt. Optional: an agent tests your live app. Free.
4
AI-FW
AI-FW sits between your AI agents and LLM backends, inspecting every prompt and response — blocking prompt injection and jailbreaks, masking PII, enforcing policy, routing between models, and keeping a full audit trail. Self-hosted. Free Community Edition with 100M tokens/month.
1
Ganymede
Ganymede 1.1 is a free LLM for reviewing code and finding security bugs. It reads up to 1M tokens, so a whole fits in one prompt. Use it in the browser or through an OpenAI-compatible API in tools like Cline, Continue, opencode or Kilo Code.
1
MCP Census – a live change feed for 48,797 MCP servers
Look up health and ownership for 48,000+ MCP servers, or audit your whole MCP config in one call. Every claim links to public evidence. Free 10/day. REST and MCP for agents.
—
Ombuto Code
Coding agents write code fast, but someone still has to prompt, wait and check. Ombuto Code takes you out of that loop. It is an open-source desktop workbench that lives in your repo. Write a PRD, architecture and tickets with an agent, each ticket with acceptance criteria. Switch on the scheduler and agents build, test and fix in a loop while an evaluation agent checks the result against the criteria. Review the report, run summary and diff. Works with Claude Code, Codex and Kimi. Apache 2.0.
2
Lex Extract
What makes Lex Extract stand out: 1. 🔒 True Zero-Access Privacy: Unlike many AI tools that log data, our unique security architecture ensures nobody—not even us as developers—can access your documents. 2. 📊 Actionable Excel Reports: We don’t just summarize; we export structured Excel & high-risk reports linked directly to the exact contract page and language. 3. 🛠️ Solo-Built via Vibe Coding: Built entirely by a non-coder using autonomous AI agents.
1
MemTether
Cross-client AI memory hub with bi-temporal governance, supersession chains, hash-chain audit, and 9 MCP tools. One SQLite shared by all agents via file-level pointers. Local-first, zero cloud.
1
Base Web3 M2M Hub
Live on-chain data for Base Mainnet — gas, DEX arbitrage, whale tracking, Aave flash loans & liquidation, token security, contract profiling. Pay per call in USDC via x402. No API key, no subscription. MCP server included.
1
No matching projects
Try another keyword or category.