A full-stack AI Red Teaming platform securing AI ecosystems via Agent Scan, Skills Scan, MCP scan, AI Infra scan and LLM jailbreak evaluation.
Security projects
Loading…
Security projects
Public projects with traces of AI coding tools in their repositories, sorted by newest listing.
Security · 978 projectsReset filters
Guardian Protocol is an open, vendor-neutral defense network for AI agents: a shared registry of known attacks mapped to MITRE ATLAS, a reference sensor that blocks them, and a sharing hub so one orga
EMILIA gives AI the rules every new hire gets, and checks them at a gate the AI can’t go around. Every payment it allows gets a signed receipt your accountant can check offline, without asking the AI
Koschei Web3 is a security validation and risk-intelligence platform that turns observable chain state, authority and evidence into accountable security decisions.
Zones, VLANs, assets, flows, exposure and vulnerabilities, in layers. Self-hosted, AGPL-3.0.
Translation pending
The Python Risk Identification Tool for generative AI (PyRIT) is an open source framework built to empower security professionals and engineers to proactively identify risks in generative AI systems.
Attack your AI agent with prompt injection and tool abuse, judge it from the tool-call log, and decide whether to approve it. Mapped to MITRE ATLAS and ATT&CK.
Data and context make agents useful. hoop.dev makes it safe. Attach a sidecar to a database, cluster, or agent sandbox and get masking, guardrails, and approvals on live traffic. No code changes, no n
Manage your WireGuard VPN with ease
Autonomous AI security testing for developers and security teams: find vulnerabilities, prove each with a working exploit, and ship verified fixes with zero infrastructure setup.
Private service names, local DNS and blocklist filtering for your network, with optional replication and sealed backups.
Mask personal data and secrets before they reach an LLM, and put them back on the way out. Local detection, reversible in-memory vault.
The trust layer between AIs and humans: signed, domain-verified information from businesses and real people, in a public log anyone can verify. Free, with open MCP, A2A and API.
dumbtv takes the ads, screen tracking and update nags out of your Fire TV over ADB. No root, none of your apps touched, every step can be undone. Free and open source.
AI-powered code review, SAST, DAST, and penetration testing — fully automated in cloud sandboxes. Proof-backed security findings for every vulnerability.
Open source defense against lateral movement across Kubernetes and AWS. Agentshield maps how your workloads really talk, investigates change with evidence, and proposes fixes you approve.
Use Microsoft Entra with OAuth 2.0 and OpenID Connect (OIDC) to protect the apps and web APIs you build. Learn how to sign in users and manage their access through our quickstarts, tutorials, code sam
Bitcoin investigation tools in your browser: follow the money, label wallets, flag patterns with reasons and write a verifiable report. Nothing to install, nothing stored on a server.
Records everything locally, runs AI on your GPU, never talks to a cloud.
Translation pending
Scans binaries, containers, live TLS and cloud KMS for cryptography, grades post-quantum risk against NCSC 2028/2031/2035, and signs the evidence with ML-DSA. Validated against 25,586 real ELF binarie
Zero-trust trade execution for Solana AI agents and trading bots: DEX routes never reach your wallet, every trade is verified before signing, and it settles at your minimum or reverts.
310 published CVE and GitHub advisories where code written with Claude, GitHub Copilot, Cursor and other AI coding tools introduced or exposed the vulnerability, each traced to the responsible commit.
AVE is the behavioral classification standard for agentic AI components. Stable IDs, AIVSS scores, and behavioral fingerprints for skill files, MCP servers, and agent plugins — mapped to OWASP, MITRE
Run any command through a local, single-run proxy and see where its traffic went, which fields it carried, whether any of it is about you, and what it still sends after you opt out.
Read-only connectors stitch humans, service accounts, keys and AI agents into one identity graph: delegation chains, blast radius, 27 deterministic detectors and a CycloneDX 1.6 Agent-BOM. Proposes di
SecurePilot helps Canadian MSPs and IT consultants assess external security posture, evaluate Loi 25 & PIPEDA readiness, turn gaps into remediation tasks, and deliver client-ready reports — from one w
Self-hosted open-source software that records signed provenance at every data handoff, so recipients can verify the recorded path in their own environment.
Akto provides real-time MCP and AI agent discovery, AI agent security testing, red teaming, agentic posture management, and guardrails. Recognized by Gartner™ as a representative AI agent security pla
Translation pending
A guard for your computer. Claude, ChatGPT, Claude Code, OpenClaw and Hermes reach your browser and files only under your rules: allow, deny or ask you first.
MarkMyAss shows exactly what it removed — and what it cannot verify. Free, open source, no account, independently verified.
CapMonster Cloud is a fast CAPTCHA solver and automatic solving service. This AI-powered tool supports reCAPTCHA, GeeTest, Cloudflare, and more. Scale the online solving of simple and complex CAPTCHAs
HearthLink connects your Windows computers from any network, encrypted end to end. You approve every session and stay in control.
It is not a browser. It sits between an agent and whatever fetcher you already run, decides each destination against your policy before anything leaves, decides every redirect again, bounds what comes
Self-hosted forward proxy for a home, lab or small-office network: a Squid forward proxy, a Go WAF over ICAP, a dnsmasq resolver that sinkholes blocked domains, a Go backend with a REST API, and a Rea
Understand and control tracking in your apps with open source privacy tools.
Scan any website and get evidence of trackers firing before consent, cookies dropped without permission, and personal data leaking to ad platforms.
Translation pending
ztd turns AI actions into governed, signed, auditable transactions instead of letting an agent speak directly to the host.
An open-source framework for detecting, redacting, masking, and anonymizing sensitive data (PII) across text, images, and structured data. Supports NLP, pattern matching, and customizable pipelines.
As enterprises integrate AI agents into their technology stacks, these agents increasingly rely on external tools and services to navigate complex
The best IP Toolbox. Check your IP address & geolocation, test IP for WebRTC and DNS IP leaks, run an IP quality check, browser fingerprint check, website availability check, network speed test, globa
Is your site affected by PHP2Shell? A free passive check of the version headers your server already sends.
Every app needs an alibi. A history of what AI agents and apps did on your Mac.
A cybersecurity analyst's coverage map across identity, corporate IT, application security, product security, and neighboring domains. Identity is the evidenced slice. Unknown stays unknown.
Quin finds the AI agents in your codebase with no instrumentation: extract system prompts, analyze each agent's intent and risk signals, and export compliance-ready reports.
Check your website's HTTPS, security headers, exposed files and mail domain settings, with fixes for your own server.
An agent's log is only as honest as whoever holds the disk. Provenant makes it evidence.
See if a pay-per-call AI service or MCP server is up, how fast it answers, and whether its price and payment address match its listing, before your agent pays. 25 free checks a day, then $0.001.
Microsoft Sentinel로 더 빠르고 간소화된 SOC를 운영하세요. 위협을 더 빠르게 탐지하고, 대응을 자동화하며, 데이터 레이크로 SIEM 비용을 낮추는 클라우드 네이티브 SIEM입니다.
Translation pending
CVE-2026-109618 / CVE-2026-109619 노출을 비파괴로 점검. 내 사이트 URL로 바로 확인.
Translation pending
Test your Web Application Firewall against common attack payloads (SQLi, XSS, SSRF, etc.) using various HTTP methods.
A binary static analysis tool that provides security and correctness results for Windows Portable Executable and *nix ELF binary formats
Bowser is an open-source browser for 64-bit Windows. Stupidly lightweight, stupidly fast, private, and every ad is gone. It runs your Chrome extensions from their folders.
免费开源的 Claude 访问环境检测。查看 Claude 出口 IP 与支持地区、16 次 DNS 泄露采样、WebRTC 泄露、IP 类型，以及与 IPPure 同算法的中文环境评分。
Translation pending
Kumo discovers what is exposed. Kage investigates what happened. Two open-source security tools, one lifecycle.
Free scanner for crypto trading-bot repos, npm/PyPI packages and Solidity contracts. Finds wallet drainers, key stealers, typosquats and fake-star repos, and runs contracts in a sandbox EVM to see whe
Scan and remove photo metadata in your browser. Photos stay on your device. Supports JPEG, PNG, WebP, GIF, HEIC, AVIF, and TIFF. RAW is part of ClearFrame Pro.
Find the AI tools your clients' staff connected to company systems, and lock down the risky ones.
Private data, untrusted input, a way out: koto keeps the lethal trifecta apart with three tiers and one microVM per agent group.
OKed sends your AI agent's risky actions to your phone. Safe actions just run. Tap yes or no, and your agents keep working. Free iPhone beta.
Security intent, embedded in code. Automate your threat model with GAL and AI.
We make tools that keep you in the loop with AI. They check what the AI is allowed to do, and catch what it missed.
Slipmark makes AI agents prove what they checked. Every verdict is bound to a SHA-256 hash of the files the agent read. Open source GitHub Action, Apache-2.0, with Slipmark Cloud for records, rules an
What an agent's tool may read, write, run and connect to, named by one hash and enforced by the kernel. Open source, Python standard library only. A real Linux kernel boots in your browser to show it.
Open-source network deception platform. Deploy silent decoy services to detect lateral movement, ransomware, and insider threats — before real damage is done.
AgentDossier — discover AI agents, inspect the evidence behind each one, and decide what fits your risk requirements.
Know every piece of software you run. Fix what matters first. Open source and self-hosted.
Models the structure of an authorised password corpus — for policy audits, honeyword decoys and synthetic data. Defensive research; authorised corpora only.
Censgate builds privacy infrastructure for AI systems. Redact, our open-core engine, detects and anonymizes 61 classes of PII and secrets in AI prompts and model responses — Apache-2.0, self-hosted, a
Scan your Claude Code history for the commands you would never have approved. Read-only, local, zero dependencies.
Plug in at home, pair with your mobile device, build and share apps privately. Not even we can eavesdrop.
Open-source governance for AI agents: runtime spend control, policy, memory, identity, cryptography, quality and pre-production drills. Cloud security, IAM and FinOps for AWS and GCP.
Authorize AI agent actions with signed policies, single-use permits and verifiable receipts. Open-source Python and Rust, with native MCP support.
Every AI capability, independently audited. Public, open-source trust scoring for skills, MCP servers, hooks, plugins, and rules across every agent platform.
Free and open source. Cloakroom masks API keys, secrets and personal data on your own machine before your AI coding agent sees them. Works with Claude Code, Codex CLI and Gemini CLI.
Open-source, self-hosted disposable email delivered as a Discord DM, Telegram message, or Slack DM. Runs on Cloudflare Workers, no account, no web UI, no third party in the loop.
TrailSift for Mac clears the record of where you've been that other cleaners miss: the download log, history kept in odd places, remembered Wi-Fi networks, cookies and old iPhone backups, with what de
Email and identity for AI agents. Source available, written in Rust, runs on your own Cloudflare account or on Pylota Mail Cloud.
Mask personal data and cut AI token costs on your device — before anything reaches Claude, ChatGPT, Gemini or any AI. Free Mac app, Chrome & Python.
Replace complex Borg Backup terminal commands with a beautiful web UI. Create, schedule, and restore backups with just a few clicks.
A free, open-source SIEM for Windows and Linux that is easy to set up, easy to query and ready for AI.
Map every path from your app's code to its Supabase database, and catch what's exposed but never used.
Paste-in tools that run entirely in your browser — decode JWTs, lint Dockerfiles, audit security headers and scan for leaked secrets. Nothing is uploaded.
Open-source security operations where AI agents investigate but never act: evidence graph, detections and an approval-gated action broker with rollback.
Open source CLI that plants decoy credentials in your project, then blocks the tool call the moment an AI coding agent reads one or carries the value off the machine. Works with Claude Code, Cursor, C
MCP tool locks, a GitHub Action and signed receipts for teams using Claude Code and Cursor.
Антидетект-браузер із власним ядром Chromium. Локальні профілі, проксі, синхронізація вікон. Free назавжди — 5 профілів.
Translation pending
Evidence-first vendor and AI tool risk review app. Python, Flask, explainable rules, live demo.
End-to-end encrypted logins, API keys and SSH keys for small teams and their agents. Open source, self-hosted on your own Cloudflare in one click.
3B+ Android users deserve access to a safer and open Internet. RethinkDNS is a private, secure, and fast DNS resolver with custom rules, blocklists, and analytics that lets you block websites temporar
The open-source CTF platform: run capture-the-flag competitions with real-time scoreboards, automations, team collaboration and one-command self-hosting.
Zero-to-secure in minutes. TLS, auth, rate limiting, WAF, and AI-readable logs. One binary, one config file.
Translation pending
The open-source CLI that sandboxes AI agents with filesystem, network, and secret fencing.
Encrypt & split your secrets into QR codes using Shamir's Secret Sharing. Zero-knowledge crypto inheritance for Bitcoin seed phrases.
Free, honest tools made by one person: fix PDFs, make strong passwords, check suspicious links, read text from photos. Everything works on your device — no ads, no tracking, no paywall at the end.
Robocall mitigation software that runs beside your switch. Verifies STIR/SHAKEN, blocks by signer, checks FCC standing, and answers tracebacks. Free and self-hosted.
ficta Gateway is a self-hosted AI workspace that lets teams review registered and detected values before model egress, add anything missed, and replace protected text with local surrogates.
Inspect what agents read and guard what they run — locally, with no model API.
Browser-based web exploitation challenge platform powered by WASM
Translation pending
Paste or upload a Cisco IOS / ASA, Cisco NX-OS, Juniper Junos, Palo Alto PAN-OS, Linux iptables, or Arista EOS config and get a ranked list of dead, shadowed, redundant, and dangerously-permissive rul
Translation pending
A free, read-only check for your Mac. Prints paths and counts, never a key.
Argus spots when many wallets paying an x402 seller belong to one operator, so promos go to real customers. Built on Monad.
Translation pending
Enterprise consent and preference management that collects, syncs, and honors customer choices across every channel, system, and data use.
Open-source trust & safety for community builders. MIT.
Fedora Atomic with KDE Plasma, hardened by default and explained in plain words, with Kali's hacking tools in a lab of their own.
Proteggi i tuoi dati offline con la potente suite AI open-source per Pubbliche Amministrazioni e aziende.
Translation pending
Free security check for AI-built apps (Lovable, Replit, Bolt, v0, Cursor). No sign-up, no login. Paste a public GitHub link for a plain-English graded report.
Protege archivos con una clave. Funciona sin internet y nada sale de tu equipo.
Translation pending
SealMyApp checks apps built with Lovable, Bolt, Cursor, Replit and v0 for open databases, leaked keys and exposed files, then gives you copy-paste fixes for your builder and keeps watching.
STIG AMI and CIS hardened AMI images for AWS. Published OpenSCAP scores on x86_64 and arm64.
Open-source network discovery, observability, camera and WiFi workflows, and edge lifecycle management.
Hand context between agents — files, prompts and decisions — with a trust boundary that survives the hop.
A video-meeting authentication portal supporting both 8x8 JaaS and self-hosted Jitsi Meet. Multiple accounts / roles / 2FA, full audit logging, scheduling and email invitations — self-hostable and ful
Translation pending
Fournisseur Conforme — auto-diagnostic cybersécurité pour fournisseurs et prestataires : évaluez vos pratiques face au projet de référentiel ReCyF v2.5 (version de travail ANSSI) et repartez avec des
Translation pending
Framework-agnostic security middleware for Node.js and edge runtimes
DEVADARA is an offline Windows antivirus that defends your computer the way your body defends itself.
Your leases, contracts and invoices, read and checked by Yomizaka: every finding quotes its rule and the line it rests on, every PDF examined for signs of editing, every document found again in a seco
QtPass is a free, open-source GUI for pass, the Unix password manager: GnuPG-encrypted passwords synced with Git on Linux, Windows, macOS and BSD.
Optics maps every AI process and every connection it makes, flags the first time AI reaches somewhere new, and sends it to the monitoring tools your team already watches. It sees the connection, not t
VEXORA (Explainable Fund-Flow Tracing for Fraud Detection): follows funds within a bank, flags potentially suspicious transfers and builds one case for the investigator. All data synthetic.
Vettalon 用多個 AI 模型交叉覆核程式碼風險，保留候選與反證供人工判斷。掃描由本機編排；深掃會將選取的程式碼與上下文送到你設定的模型。免費核心提供供應鏈獵捕，Pro 提供企業報告與桌面 App。
Translation pending
Quantsiv finds the cryptography in your code that a future quantum computer could break, shows what to fix first based on how long your data must stay secret, and produces the Cryptographic Bill of Ma
Encrypted legacy vault. Store messages, passwords and documents, pass them to people you choose only when it's truly time.
Protect your Node.js projects from supply chain attacks. Scan npm packages for malware, crypto-drainers, and security vulnerabilities with AI-powered threat intelligence.
Self-hosted OAuth 2.1 for your apps, APIs, and AI agents. On-behalf-of delegation, MCP support, and a crypto-agile core built for post-quantum security.
One-time keys for an ordinary Safe: a seat that keeps its place among the owners while the key behind it changes after every approval. Your passkey plays the key. Base Sepolia only.
Secure DevOps workflows with Smallstep. Automate trust and integrate device identity across DevOps pipelines for enhanced security.
An open specification and tools for verifying the provenance of scientific detector data, including images, movies, and spectra, using signatures made inside the detector.
RouteGate is an open-source self-hosted platform for managed VPN nodes, accounts, devices, routing, and client access.
Encrypted in your browser, stored as an Arkiv entity, deleted by the protocol. The key never leaves the link.
ProvGuard: provenance-based tool-call guard for LLM agents, evaluated on AgentDojo.
The OWASP OFFAT tool autonomously assesses your API for prevalent vulnerabilities, though full compatibility with OAS v3 is pending. The project remains a work in progress, continuously evolving towar
Class action settlements, tracking protection, automatic cookie banner handling and global VPN locations in one app.
Translation pending
Record every host an MCP server contacts, per tool, and fail CI on new ones.
AI-native cybersecurity maturity for SMEs and the consultants who serve them.
Manage ISO 27001 compliance end-to-end. Gap tracking, risk register, evidence automation, and 29 regulatory framework assessments — self-hosted, open source. Multi-jurisdiction policy rendering for 8
Surf the internet from home, anywhere in the world! Fully managed, Zero trust VPN for your home or small office network.
Let Claude Code use your passwords and API keys without ever being handed them. Free, open source, Windows.
Live CVE tracking across leading cybersecurity vendors — Fortinet, Palo Alto Networks, Cisco, Check Point and more — broken down by firewall, endpoint, SASE and identity products.
Translation pending
Run Claude Code, Codex or pi in a secure Linux VM on your Mac. Your API keys never enter the sandbox, every connection follows your policy, and it pauses in 1 ms and wakes in 0.3 s.
AEGIS Defense: sistema autónomo de ciberdefensa post-cuántica desplegado en producción. No es una librería. No es un concepto. Es defensa activa 24/7.
Translation pending
See what your Microsoft 365 tenants over-share, how their posture scores and who can spoof their domains, for one environment or a whole client book.
One package, one sentence to your AI agent, about 20 minutes. Compass runs entirely in your AWS account.
Translation pending
Slopscale is a fork of headscale, the self-hosted Tailscale control server, that runs one tailnet with an admin console, roles, approval and auditing.
Keeps the Internet open and your data secure. Top-notch VPN from the creators of the famous ad blocker.
Get SOC 2, GDPR, HIPAA, and ISO certified with expert guidance and the job done for you. Open-source compliance platform with no vendor lock-in.
Hybrid Rust/Python security scanners for prompt injection, DLP, PII, and agentic tool risks — a layered, on-device AI firewall library.
API tokens for Blablador, issued after Helmholtz/JSC sign-in
Translation pending
One static agent for Windows, macOS and Linux. Inventory, posture, compliance, policies, deployment and remote actions, with a public roadmap to native MDM.
An open-source, large-scale, full packet capturing, indexing, and database system. Stop using slow tools to dissect and search your packets, let Arkime do the grunt work for you!
From shadow AI to signed evidence. Discover, sign, and hand the auditor a bundle they can verify themselves.
A person signs the permission with a passkey. The agent signs a receipt for each action. A free checker says whether the agent stayed within the permission, and where it did not.
Your AI agent says it did the right thing. Deedglass checks if it really did, and stops it if it didn't. Free to run on your own servers.
Checks AI-generated code before it merges: security rules, SARIF output for GitHub code scanning, and a CI gate that exits 1 on HIGH findings. Open-source core (MIT).
Discover your AI agent's attack surface — capabilities, tool permissions, MCP integrations, and prompt risks — with clear, actionable findings before deployment. Free, offline, open-source.
Corgea finds, triages, and fixes vulnerabilities across code, packages, infrastructure, and containers.
Translation pending
Trust that travels with your data. Every signature, AI action, and file carries its own proof — so anyone who asks can verify what happened, anywhere.
A modern GUI client based on Tauri, designed to run in Windows, macOS and Linux for tailored proxy experience
Protect your SIP Servers from bad actors at https://sentrypeer.org
Secrets Bridge adds access requests, approvals, audit and agent execution around the secrets you keep in AWS Secrets Manager and HashiCorp Vault. Values stay in your provider.
A self-hosted gateway that checks, limits, masks and audits every statement before it reaches your production database.
One goal becomes a governed, audited plan. Your deployment's rules can veto it; the record survives root access. Apache 2.0.
Remove common tracking parameters from URLs locally in your browser.
A free, native Mac app for API keys, .env variables and recovery codes. Encrypted on your Mac and opened with Touch ID.
One audited Postgres base mapped to SOC 2, HIPAA, ISO 27001, and NIST 800-53: fail-closed RLS, S3 Object-Lock WORM, and an append-only audit chain, in five composable module families you compose, neve
Privacy-first apps for Mac and iPhone. A Faraday cage for your data.
A live-fire range where anyone can try to rob an AI agent's wallet. Breach it and the contract pays you; fail and the agent's on-chain score — and spending limit — goes up. Settled in USDC on Arc.
Open security scanner and self-hosted control plane for AI, MCP, and cloud infrastructure
Claude Code, Codex and Cursor ask Paveo before they run a shell command or write a file. The dangerous ones are refused, with a record your security team can verify.
Screen names against signed sanctions data in your browser. No account or backend required.
Prompt injection detection for autonomous agents. Seven technique classes. Sub-10ms response. Open source.
Не работает Telegram, WhatsApp, YouTube? Бесплатный прокси за 10 секунд и VPN для обхода глушилок и белых списков. МТС, МегаФон, Tele2, Билайн, Yota.
Translation pending
Three open-source layers: fence the OS with NockLock, firewall the tools with NockGuard, view configured audit records on the Wall.
Turn sensitive report snips into client-safe replicas. Sensitive text is masked on your device. Nothing is stored.
Turn online abuse into a case you can file. Check images for manipulation, keep tamper-evident evidence, draft your NCCIA complaint and find the right helpline — in English and Roman Urdu. Your data s
Your AI writes the code. Kompile makes sure it does not leak. Check your app for exposed keys and open doors, fix them with your own AI, and stay protected every week.
Decode a SAML response or OIDC token, run the hard checks, and get the fix plus two ready replies.
Third-party risk review in minutes: tier the vendor, check its SOC 2 evidence, and get a decision and memo.
A spend firewall for AI agents: every payment is checked against the job before money moves. Right payee, right amount, in scope, plus a verified-payee credential read — one call returns GO or STOP wi
Detect attackers while they're still mapping your APIs. trappsec embeds decoy routes and honey fields directly in your application code — catching adversaries before a single perimeter is crossed.
Receive your SMS online with a virtual number: create a WhatsApp, TikTok or Telegram account without a SIM. Instant SMS validation, Mobile Money payment in FCFA.
AI translation
Policy-gated before it runs, cryptographically provable afterwards. Check a real log in your browser.
Translation pending
Privacy-preserving, client-side AWS IAM policy analyzer. See a single policy's potential blast radius - privilege-escalation paths, role-assumption reach, sensitive-data access, and broad-resource gra
Translation pending
Built it? Time to show it.
One command in your project folder. Built with AI is enough; AI features are optional.
/nomorevibeSee how to list