Open-source tools for Security & privacy

Security scanning, privacy, evaluation and protective tooling.

Look at what the tool checks and the evidence it returns. Try it against a sample with a known issue to see how it explains its findings.

54 projects

agent-2 aaravshah1311 · MIT
A self-hosted AI agent that runs on your own machine with your own Gemini API key, combining a coding assistant, terminal agent, and security tester with persistent…
agent-safe-pipeline decionis · Apache-2.0
A blueprint for keeping AI agents from taking real-world actions on their own, by forcing every action through an independent policy check and a human approval step.
agentplay entelligence-ai · Apache-2.0
Turns a transcript of your AI coding-agent session into a visual map and timeline you can replay.
ai-infra-guard tencent · Apache-2.0
Tencent built an open-source security scanner that probes AI agents, agent skills, and LLMs for jailbreaks and other vulnerabilities before attackers do.
anthropic-cybersecurity-skills mukul975 · Apache-2.0
Give AI coding agents structured cybersecurity skills mapped to established security frameworks.
argus thomasunise · MIT
An open-source platform where AI agents watch your cloud, code, and network for attacks, investigate them, and can contain them in seconds while a human stays in control.
bla qybaihe · MIT
BLA protects shipped apps (Electron, JS, Python, WASM, native binaries) from AI reverse-engineering agents by making the analysis burn tokens and follow decoys instead…
box strands-agents · Apache-2.0
Box runs your AI coding agent inside a locked-down sandbox so it can only touch the files, commands, and network endpoints you explicitly allow.
bug-bounty-hunting-prompts mdpsec · MIT
A free, open-source set of LLM prompt files that turn bug bounty hunting into a structured, evidence-first workflow instead of ad-hoc poking around.
claude-plugins-community anthropics · Apache-2.0
Anthropic's own community plugin marketplace for Claude Code and Claude Cowork: a read-only mirror listing reviewed, security-scanned community plugins you can install…
codex-security openai · Apache-2.0
An AI-agent-powered CLI from OpenAI that scans code for security vulnerabilities, validates whether they are real, and can fix them automatically.
ecc affaan-m · MIT
An open-source performance layer for AI coding agents like Claude Code, Codex and Cursor, adding memory, skills and security so agents keep context and work more…
echomind nirbhay41120003 · MIT
A local-first assistant that transcribes your speech live and lets you ask a private RAG chatbot questions grounded in what you have actually said, entirely on your…
eleven zfinix · Apache-2.0
An app that lets you drop in an app, binary, Android APK or GPU kernel and have an AI agent explain what it does, rename functions and suggest patches, while you…
geiger atomburstofficial · MIT
A single command that scans your machine and tells you, in plain language, every AI agent, plugin, and MCP server installed and what each one can touch.
grim-prompter jmanico · MIT
Grim Prompter turns an AI coding agent into a threat-modeling partner, walking 36 architecture patterns and 29 security methods and returning evidence-traced findings.
hello-agent-system heaven999b · MIT
A free, hands-on 17-lesson course that walks you through building a production AI agent system from scratch, covering tools, reliability, security, evals, and cost.
hob frigadehq · MIT
A free, open-source desktop app that lets anyone chat with AI models on their own computer, offline and privately, as an alternative to ChatGPT/Claude.
ifixai ifixai-ai · Apache-2.0
An open-source auditing tool that checks whether an AI agent is actually doing its job correctly, not just technically functioning, giving a pass/fail scorecard in…
imagegen-mac ph1lb4 · MIT
A native Mac app that runs an image-generation AI model fully on your own device, so you can create and edit images offline with no subscription, no API key, and…
Jōb youssefamr74 · MIT
Jōb is useful for anyone in the middle of a job search who’s tired of tracking applications in a spreadsheet, especially people applying to a lot of roles at once. It…
kancelaria-lex ai4charitypl · MIT
An open-source AI tool for Polish lawyers that reads case files and answers questions with citations traced to the exact source sentence, engineered so it cannot invent…
keelflow perruer · Apache-2.0
A self-hosted visual builder for AI agents and LLM workflows, forked from the abandoned Flowise project with known vulnerabilities patched and all commercial-license…
kensaku yjid · MIT
A free, open-source file launcher that lets you instantly search inside documents, code, and screenshots, using OCR and semantic search, entirely on your own machine.
kokon gabrielelozupone · AGPL-3.0
A free app for Windows and Mac that lets you edit a photo just by describing the change you want, running fully offline on your own GPU with no cloud upload.
kru omaekumiko2-create · MIT
KRU is a local, open-source credential vault that lets AI coding agents use your passwords, API keys, SSH keys, and 2FA codes to do their job without ever seeing the…
llm-hallucination-detector syasj · MIT
Color-codes an AI chat answer token by token based on model confidence, and separately checks each factual claim against your own source documents, flagging guesses vs.…
maskit xiayutian11 · AGPL-3.0
A local proxy that sits between your AI coding tool (Cursor, Claude Code, Codex) and the model provider, automatically masking secrets and personal data in outgoing…
northcinder cinderline · MIT
NorthCinder is a self-hosted MCP server that has an AI agent compare and rank products for you, show its reasoning and rejected options, and require your sign-off via a…
OG Converter dr-at0mic · GPL-3.0-or-later
Office workers can turn Word, Excel and PowerPoint files into PDFs, merge or split PDFs, and password-protect documents without uploading them to a website. Students…
onlyread onlyreadai · AGPL-3.0
A Chrome extension that lets an AI browser agent read everything it needs across your real accounts, while blocking every send, post, edit or delete before it ever…
open-code-review alibaba · Apache-2.0
An open-source code review tool from Alibaba that combines rule-based checks with an LLM agent to leave precise, line-level review comments and catch bugs and security…
openhuman tinyhumansai · GPL-3.0
OpenHuman is an open-source personal AI assistant that builds a local, private memory of your life and work, then uses that memory to research, automate tasks, and…
openqodex openqodex · Apache-2.0
A free CLI that runs security/lint scanners on just your changed lines, then has a separate Claude Code or Codex process review every finding against the real diff…
openwhispr openwhispr · MIT
OpenWhispr is a free, open-source, privacy-first alternative to WisprFlow and Granola: press a hotkey, speak, and your words appear at your cursor, with fully offline…
pentagi vxcontrol · MIT
An open-source AI agent system that autonomously plans and runs penetration tests inside an isolated Docker sandbox, using a team of specialized LLM agents and 20+…
pentest-harness s1n6h · MIT
A self-hosted AI agent harness that helps run authorized penetration tests, bug-bounty work and CTF challenges, keeping every session local to your machine.
personal_external_brain chenkanglin198904 · MIT
A local-first tool that turns your notes, PDFs, web pages and audio/video into a private knowledge graph, and is designed to say I do not know rather than guess.
procoder azrtydxb · Apache-2.0
Procoder gives AI coding agents a real quality gate: a single Go binary that blocks commits with unformatted code, merge-conflict markers, or junk files, and turns…
rea morluto · MIT
Lets your AI coding agent reverse-engineer a feature you like in any app, explain exactly how it works with evidence, and rebuild a version of it for your own product,…
rea-video juneyaooo · Apache-2.0
A tool that watches any video and, shot by shot, tells you whether it was filmed, AI-generated, or built with code/animation, then hands you the prompts and recipe to…
replica-skill jakeschincariol · MIT
Eleven free Claude skills that work as a team to reverse-engineer any app, rebuild it clean-room style (features and flows only, never code/logo/content), test it for…
repoward sanoy24 · MIT
It's for developers and teams who let AI coding agents (Claude Code, OpenCode, Cursor) work on their GitHub repositories. With it they can give an agent GitHub access…
reverify 2akouwu · MIT
Reverify pairs an AI model with deterministic binary-analysis tools that check every claim against the real file, so reverse-engineering output is grounded in the…
ruview ruvnet · MIT
Open-source project that uses WiFi signals and on-device ML pose estimation to detect presence, movement, and vital signs, no camera needed.
sapper-v2 n1neman · MIT
An open-source multi-agent AI system that autonomously finds and exploits vulnerabilities in a competition setting, coordinating a commander agent and worker swarm.
security-audit-skill cloudflare · MIT
Cloudflare open-sourced a skill that turns a coding AI agent into a security auditor, running it through recon, bug hunting, and independent verification before…
skillspector nvidia · Apache-2.0
Scans AI agent skills for Claude Code, Codex, and MCP for vulnerabilities and malicious patterns before you install them, after finding that over a quarter of a large…
sparky anisayari · MIT
A free, open-source private AI assistant for Windows that runs entirely offline: chat like ChatGPT, see images, generate pictures and videos with sound, search your own…
universal-modder rehan-remade · MIT
Point an AI coding agent (Claude Code, Codex, Cursor, Gemini CLI) at a PC game you own and it mods it: finds the engine, reads the code, generates new art/3D/audio with…
websitebench tiger-ai-lab · Apache-2.0
A benchmark that tests whether AI agents can rebuild a real website just by exploring it in a browser.
x64dbg-mcp-server duty1g · MIT
Lets any MCP-compatible AI assistant control the x64dbg debugger programmatically over HTTP.
xiryn souravberaakaspeed · MIT
Lets you run your own open-source version of a personal AI agent (like Metas Muse or OpenAIs Dots) entirely on your own machine: each agent gets a private sandboxed…
yogurt frigadehq · MIT
A free, open-source desktop app that records your meetings, transcribes them locally with whisper.cpp, and writes AI summaries with a local model, no bot added to the…