剑庭安
首页
黑客安全首页全部安全栏目 黑客技术黑客技术、渗透测试、漏洞利用工具与教程 安全软件安全工具、扫描器、检测与审计软件 病毒分析恶意软件分析、样本、逆向与沙箱 CTF思路CTF 题目、writeup、解题思路与工具 防护思路防御技术、加固、检测响应与蓝队 红蓝攻防红队攻击模拟、蓝队对抗、攻防演练 漏洞知识库漏洞、CVE、PoC、EXP 与安全研究
逆向|反编译
PHP源码PHP 网站源码、程序、系统 APP源码安卓/iOS APP 源码 ASP源码ASP 网站源码与程序 NET源码.NET 网站源码与系统 小程序源码微信/支付宝小程序源码 Dedecms风格模板Dedecms 织梦风格模板 Empirecms风格模板EmpireCMS 帝国风格模板 Discuz商业模板/插件Discuz 商业模板与插件 WordPress主题/插件WordPress 主题与插件 其它源码、模板其它各类源码与模板 游戏源码游戏源码与引擎项目 加密货币源码加密货币、区块链、钱包、USDT/比特币相关源码与工具 手游源码手机游戏源码、服务端、客户端与引擎项目 传奇源码热血传奇、传奇世界、复古传奇等客户端/服务端/引擎源码 页游源码网页游戏、H5 游戏、浏览器游戏源码 端游源码电脑单机游戏、PC 网游客户端/服务端源码
最新万能免驱动支持所有型号打印机下载程序 日历天气老黄历 · 每天同步真实日期与天气
站长工具站长工具 自助开通会员浏览该分类 其他资源qtziy SEO工具SEO工具 站群源码站群源码 泛程序泛程序 蜘蛛模板蜘蛛模板 蜘蛛池蜘蛛池 外链工具外链工具
搜索 自助开通会员🔓 破甲 API
登录 注册
首页 分类列表 红蓝攻防 promptzero - AI operated Fli…
promptzero - AI operated Flipper Zero 红蓝攻防 点击放大
作者:admin 发布:2026-10-06

promptzero - AI operated Flipper Zero

项目说明<p align="center"> <img src=".github/assets/banner.png" alt="PromptZero — natural-language operator for the Flip…
资源可下载 免费下载 0 条讨论
会员资源
开通会员后即可下载全部资源。
登录后开通会员
会员方案
488浏览量
0讨论
2下载链接
可下载资源状态
内容可验证
更新 2026-10-06
人工审核发布
评论区答疑

promptzero - AI operated Flipper Zero

promptzero - AI operated Flipper Zero

项目说明

<p align="center">
<img src=".github/assets/banner.png" alt="PromptZero — natural-language operator for the Flipper Zero" width="560">
</p>

<p align="center">
<a href=" src=" alt="Latest release"></a>
<a href=" src=" alt="AGPL-3.0"></a>
<a href=" src=" alt="CI"></a>
<a href="scripts/claude-usage.py"><img src=" alt="Claude usage"></a>
</p>

<sub align="center">_Claude usage/cost is an estimate aggregated from the maintainer's local Claude Code transcripts (task claude-usage); it reflects this machine's sessions only._</sub>

> Describe it. Generate it. Deploy it. Run it.

PromptZero is a natural-language operator for the Flipper Zero. Talk to it like you'd talk to a person — it generates payloads, deploys them, and runs them, all from a single sentence.

> [!CAUTION]
> Authorised use only. PromptZero generates and runs RF, NFC, RFID, and HID payloads — illegal outside contexts you own or ha authorisation to test. Read SECURITY.md for the safety model and threat boundary. The project is under acti; APIs and tools change between minor .
>
> *Built end-to-end with Claude. Review generated payloads before deployment.*

---

Quick start

Prerequisites — Flipper Zero with modded firmware (Momentum / Unleashed / RogueMaster), an Anthropic API key, and a USB cable.

That's the whole onboarding. From here, type natural-language instructions or /help to see slash commands.

> Defensi (one-flag safety rail): add --read-only to refuse any tool that writes, transmits, or executes. Pure reads / scans / queries still dispatch; anything risk-Medium or above is refused at the boundary. See Read-only safety rail for the full rule.
>
>

Other paths: Windows users grab the .zip from the releases page. WSL2 needs USB passthrough — see Transports → WSL2. For wireless BLE, config files, personas, en , and self-upgrade: Configuration reference.

---

What it does

PromptZero connects to your Flipper Zero (and optional ESP32 Marauder WiFi de) over USB serial or BLE, then lets you control e through natural language.

| Subsystem | Capabilities |
|---|---|
| Flipper Zero | Sub-GHz TX/RX, IR TX/RX, NFC detect/emulate, RFID read/write/emulate, iButton, GPIO, BadUSB, storage, app launcher |
| ESP32 Marauder | WiFi scan, deauth, beacon spam, probe flood, PMKID capture, e, BLE spam, BT scanning, skimmer detection, wardriving, MAC spoofing |
| AI Generation | E HTML, BadUSB DuckyScript, Sub-GHz .sub, IR .ir, NFC .nfc from natural language — plus parametric builders for typed parameters |
| Intelligence | Image analysis (file path → device ID + attack ), SD card discovery |
| Audit | SQLite audit log with MITRE ATT&CK technique tags, session export, statistics |

Run promptzero and type /tools for the li. Tool count grows release-o.

The agent layer ships prompt caching, cost-tier model routing (recon on Haiku / exploit on Opus), prompt-injection quarantine, reflexion-on-error with structured ToolError, a <de> oracle injected each turn, and OpenTelemetry GenAI spans. See docs/ for architecture details.

---

Modes

CLI — promptzero

Default. Interactive REPL.

Slash commands (run /help for the full list with descriptions):

  • Con: /help, /reset, /quit

  • Session: /sessions, /save <name>, /resume <id>, /forget <id>

  • Info: /status, /tools [query|page <n>], /history [N], /audit {stats|find|tail|top|session|query|export}, /stats [section], /cost, /budget [set <USD>|off], /debug

  • Operator: /persona [name], /mode [name], /watch [pause|resume], /webhooks [test <name>], / <path>, /attack {set|clear} <techniques>, /campaign {|run} <file>, /rewind [snapshot], /report [session] [json] [save], /rules [list|pause|resume|test]

  • Device: /reconnect
  • Keystrokes during a turn:

  • Ctrl+C — cancel the current turn entirely.

  • Ctrl+G — abort the current streaming tool (e.g. subghz_receive, wifi_scan_ap) but let the agent continue with the partial result. Use this when you've seen what you needed and don't want to wait out the full duration.

  • Ctrl+R — re history search.

  • Ctrl+L — clear screen.
  • Web UI — promptzero --web

    Dark-themed browser interface at . Includes the chat surface, a li , file browser, audit log, report builder, and (when a Marauder is connected) a TFT display panel.

    <p align="center">
    <img src=".github/assets/web.png" alt="PromptZero web UI — chat, li , file browser, audit log" width="900">
    </p>

    > [!IMPORTANT]
    > Auth. Set web.token in your config or PROMPTZERO_WEB_TOKEN in env. The browser picks up #token=… from the URL fragment on first load and caches it in sessionStorage. Empty token + non-loopback bind → ser a red warning. PromptZero speaks plain HTTP — terminate TLS at a reverse proxy (Caddy / Traefik / nginx) or a Tailscale / Cloudflare tunnel.

    Host it with Docker. Multi-arch images (distroless, signed, with SBOM + pro) are published to ghcr.io/xunholy/promptzero on every release — see docs/deploy/docker.md.

    Health probes. GET /healthz (li, 200 ok) and GET /readyz (readiness, JSON subsystem snapshot) are unauthenticated for headless / systemd / k8s deployment; device connectivity is reported but does not gate readiness (a Flipper/Marauder may attach after startup).

    Voice — promptzero --voice

    Push-to-talk in CLI mode. Press Enter with no text to record (requires sox); audio is transcribed Whisper, then processed as a normal command.

    MCP — promptzero --mcp

    Runs as a Model Context Protocol server o. Add to Claude Desktop / Claude Code:

    > [!IMPORTANT]
    > MCP risk gate. Risk-High and Risk-Critical tools are refused by default — set PROMPTZERO_MCP_ALLOW_HIGH=1 and / or PROMPTZERO_MCP_ALLOW_CRITICAL=1 to allow them. All MCP calls (allowed or denied) are recorded in the audit log. See Safety model below.

    ---

    Safety model

    PromptZero is dual-use offensi. The safety story is the project's social licence to exist.

  • Risk classification per tool. Every spec carries a tier — Low / Medium / High / Critical. Read-only ops are Low; data writes and signal emulation are Medium; active RF transmit, tag writes, and script execution are High; brute-force and destructi are Critical.

  • Consent gate. High and Critical tools require operator confirmation. The CLI shows a boxed preview (frequency / modulation / hex) with a 2-second delay; positi (y, all, confirm) is rejected before the delay opens. Negati (n, r for revise, Esc) bypass the delay.

  • MCP refuses by default. No MCP client can run High+ tools without explicit env- (see above).

  • Audit-log fail-closed. If no audit log is initialised, the agent refuses High+ actions rather than running them silently.

  • Prompt-injection quarantine. Tool outputs (scanned SSIDs, captured packets, image content, SD filenames) are wrapped before being fed back into the model so they can't o the system prompt.

  • No auto-deploy of generated payloads. BadUSB scripts deploy without execution by default.
  • Read SECURITY.md for the full threat model, scope / out-of-scope, and how to report a .

    ---

    Compatibility

    | Firmware | Status |
    |---|---|
    | Momentum (formerly Xtreme) | Primary target |
    | Unleashed | Supported |
    | RogueMaster | Supported |
    | Official (OFW) 1.x | Supported with reduced feature set — region-locked Sub-GHz, no rolling code |

    > [!NOTE]
    > Official firmware locks Sub-GHz TX to region-specific ISM bands and blocks rolling-code protocols. Modded firmware unlocks the full CC1101 range (300–348 / 387–464 / 779–928 MHz) and enables TX for all 52 supported protocols.

    ESP32 Marauder de requires firmware v1.11.1+ over USB CDC ACM (/de for the official Flipper WiFi de, baud 115200).

    For BLE wireless (no cable, all tools work, ~10× slower than USB) and per-platform pairing: see Transports reference.

    ---

    Documentation

  • docs/ — handbook with task-oriented scenarios, prompt patterns, and reproducible transcripts.

  • docs/reference/tools.md — a curated reference of the core tools' schemas, risk levels, and the prompts that fire them reliably. Run /tools (or tool_search) for the complete li.

  • docs/reference/transports.md — serial, BLE, WSL2 setup.

  • docs/reference/configuration.md — config file, env vars, personas, rules, self-upgrade.

  • SECURITY.md — threat model and disclosure policy.

  • examples/ — copy-paste templates: config, rules, four operator personas (red team / blue team / CTF / hardware lab).

  • CHANGELOG.md — what's in each release.
  • ---

    Build & contribute

    See CONTRIBUTING.md. Short :

    ---

    License

    AGPL-3.0-or-later. Hosting a modified PromptZero as a network service requires publishing source changes under the same license.

    ---

    <sub>Built with Claude by xunholy.</sub>

    项目信息

    开发语言:Go

    Star 数:243

    技术要点

    该项目属于红蓝攻防方向的开源资源,代码结构清晰,适合安全研究人员学习参考。项目围绕AI operated Flipper Zero展开,提供了可运行的实现,对理解相关安全机制、工具原理与攻防技术具有较高的研究价值。

    从技术实现角度看,此类项目的核心在于对底层原理的深入理解与工程化落地。开发者在研究过程中可以结合官方文档与社区讨论,梳理项目的初始化流程、核心模块职责与数据流转,逐步掌握其中的技术细节与设计思路。

    使用与学习建议

    1. 下载源码压缩包,解压到本地环境。

    2. 根据项目类型准备对应的运行环境与依赖。

    3. 按照 README 中的步骤完成配置与部署。

    4. 在隔离的测试环境中运行验证,避免影响生产系统。

    免责声明

    本资源来源于开源社区,仅供安全研究、技术学习与合法授权范围内的攻防演练使用,请勿用于任何非法或侵害他人权益的用途。使用本项目产生的一切后果由使用者自行承担。

    资源标题promptzero - AI operated Flipper Zero
    所属分类红蓝攻防
    发布作者admin
    发布时间2026-10-06 00:46:01
    下载权限 免费下载
    附件数量2 个下载链接
    浏览次数488
    讨论数量0
    主题状态正常讨论中

    本资源免费,登录后即可在本页「下载链接」区域获取全部附件。

    请在评论区留言说明具体链接,或联系作者补档。楼主更新后本页会同步最新的下载地址。

    请先阅读正文中的授权说明;未注明的默认仅限站内学习交流,禁止直接搬运到其他站点二次售卖。
    还没有讨论,快来抢沙发~
    登录 后才能参与讨论
    下载链接
    会员才能下载
    登录后开通会员
    同分类推荐
    暂无同类资源
    免费登录即可下载
    查看会员
    放大预览

    © 2026 剑庭安 - 网站开发定制 SEO 优化蜘蛛池代做 在线看手相算命. 保留所有权利。

    Powered by 酷像SEOv3.0

    网站地图 Sitemap.xml
    首页 全部分类 黑客安全 逆向|反编译 精品源码 站长工具自助开通会员其他资源SEO工具站群源码泛程序蜘蛛模板蜘蛛池外链工具创意乐园 搜索 会员服务 登录 注册