jeet-ganguly/birdy-edwards

## 架构 ## 功能 - 🔍 **Profile collection** — Automated data gathering of posts, photos, reels, and about data - 🧠 **Interaction intelligence** — AI sentiment, stance, emotion, and language analysis per interaction - 📊 **Actor scoring** — Weighted composite score with 5-tier classification system - 🌍 **Country detection** — LLM identifies country of origin from profile signals - 👤 **Face intelligence** — Face detection, 128D encoding, and identity clustering across all images. It uses HOG model. - 🕸️ **Network graphs** — Interactive HTML graphs including force-directed and co-interactors relationship matrix - 📄 **PDF reports** — Professional intelligence report with limited ammount of charts - 🤖 **Local AI** — Ollama powered, gemma3:4b/12b/27b and other models(mention in web panel), runs on GPU or CPU - 🐳 **Docker ready** — One command deployment on Linux and Windows ## 文档 Full usage guides for both investigation modes are available on the project website: - [Click Here](https://jeet-ganguly.github.io/profile/edwards.html) [](https://jeet-ganguly.github.io/profile/edwards.html) ## ⚠️ 免责声明 ## 系统要求 | Component | Minimum | Recommended | |---|---|---| | OS | Ubuntu 24.04 LTS / Windows 10+ | Ubuntu 24.04 LTS | | RAM | 8 GB | 16 GB | | Storage | 20 GB free | 40 GB free | | Docker | Docker Desktop / Engine | Latest stable | | Ollama | Latest | Latest | ## 安装 ### 先决条件 **Step 1 — Install Docker** - **Linux:** https://docs.docker.com/engine/install/ubuntu/ - **Windows:** https://docs.docker.com/desktop/install/windows-install/ **Step 2 — Install Ollama** - **Linux:** ``` curl -fsSL https://ollama.com/install.sh | sh ``` - **Windows:** Download installer from https://ollama.com/download **Step 3 — Start Ollama bound to all interfaces** - **Linux:** ``` OLLAMA_HOST=0.0.0.0:11434 ollama serve ``` To make this permanent: ``` sudo systemctl edit ollama ``` Add: ``` [Service] Environment="OLLAMA_HOST=0.0.0.0:11434" ``` ``` sudo systemctl daemon-reload && sudo systemctl restart ollama ``` - **Windows:** Ollama listens on all interfaces by default — no extra configuration needed. If no then, ``` $env:OLLAMA_HOST="0.0.0.0" ollama serve ``` ### 快速开始 **Step 1 — Clone the repository** ``` git clone https://github.com/jeet-ganguly/birdy-edwards.git cd birdy-edwards ``` **Step 2 — Create required files and directories** - **Linux:** ``` mkdir -p app/reports app/face_data app/post_screenshots app/status touch app/fb_cookies.pkl app/socmint.db app/socmint_manual.db app/.ollama_model ``` - **Windows (PowerShell):** ``` New-Item -ItemType Directory app/reports, app/face_data, app/post_screenshots, app/status New-Item -ItemType File app/fb_cookies.pkl, app/socmint.db, app/socmint_manual.db, app/.ollama_model ``` **Step 3 — Build the Docker image** ``` docker compose build ``` **Step 4 — Start the container** ``` docker compose up -d docker compose logs -f ``` **Step 5 — Open the web UI** ``` http://localhost:5000 ``` ### 拉取 AI 模型 Pull a model on your host machine: ``` ollama pull gemma3:4b ``` Or use the **AI Model panel** in the web UI — select a model and click **Apply & Pull**. | RAM | Recommended Model | |---|---| | 8 GB | gemma3:4b | | 16 GB | gemma3:12b | | 32 GB | gemma3:27b | ### 导入会话 Cookie BIRDY-EDWARDS requires a valid Facebook session. Use the **Cookie-Editor** browser extension — works on all platforms, no Selenium required. 1. Install Cookie-Editor → [Chrome](https://chrome.google.com/webstore/detail/cookie-editor/hlkenndednhfkekhgcdicdfddnkalmdm) · [Firefox](https://addons.mozilla.org/en-US/firefox/addon/cookie-editor/) 2. Log into your **dedicated investigation account** on Facebook 3. Click Cookie-Editor while on facebook.com 4. Click **Export → Export as JSON** 5. Go to `http://localhost:5000/tools/import-cookies` and paste ## 故障排除 **Ollama not reachable from Docker** ``` docker exec -it birdy-edwards curl http://host.docker.internal:11434/api/tags ``` If it fails, restart Ollama with `OLLAMA_HOST=0.0.0.0:11434 ollama serve` **DB error: no such table or other DB related error** Start a new investigation — schema is created automatically on first use. If you stop process/program during analysis delete that investigation and then start new investigation. **Cookies expired** Go to `http://localhost:5000/tools/import-cookies` and re-import fresh cookies. **Port 5000 already in use** Change in `docker-compose.yml`: `"5001:5000"` then access at `http://localhost:5001` **Out of memory during build** Increase Docker Desktop memory to 8 GB+ via Settings → Resources → Memory - Features that bypass platform security controls - Changes that introduce cloud dependencies - Code that stores or transmits investigation data externally ## 致谢 - Inspired by [Sherlock](https://github.com/sherlock-project/sherlock) - Inspired by the OSINT and threat intelligence research community - [SeleniumBase](https://github.com/seleniumbase/SeleniumBase) — Undetected Chrome automation - [Ollama](https://ollama.com) — Local LLM inference engine - [face_recognition](https://github.com/ageitgey/face_recognition) — Face detection and encoding library - [pyvis](https://github.com/WestHealth/pyvis) — Interactive network graph visualization - [reportlab](https://www.reportlab.com) — PDF generation - [pytesseract](https://github.com/madmaze/pytesseract) — OCR engine wrapper

标签：AI风险缓解, Docker, Flask, HOG特征, LLM评估, Ollama, PDF生成, Python, SEO: Facebook情报, SEO: 本地LLM, SEO: 社交图谱分析, SEO: 自动化情报, SOCMINT, Webhook, 个人数据收集, 五级分类, 交互智能, 人脸检测, 人脸编码, 关系矩阵, 力导向图, 加权评分, 可扩展分析, 国家检测, 威胁报告, 安全防御评估, 情感分析, 情绪分析, 无后门, 本地AI推理, 照片收集, 目标画像, 社交媒体情报, 社交网络分析, 立场分析, 线下情报, 网络图谱, 自动化情报收集, 行为画像, 视频收集, 语言分析, 请求拦截, 逆向工具, 零云依赖