How scs-labrat/oh-my-osint works
Project type and purpose
• This is a single-page, full-stack Python application built on Streamlit. It provides an interactive OSINT (Open Source INTelligence) reconnaissance assistant that:
• Presents a web UI (Streamlit)
• Automates selection and execution of CLI tools
• Queries external OSINT APIs
• Leverages Google Gemini (LLM) for tool recommendations and report generation
• Generates and serves downloadable Markdown reports
File structure
• Top-level scripts:
• install-script.sh / oh-my-osint.sh – helper scripts to bootstrap and install dependencies and CLI tools
• oh-my-osint.py – main Python orchestration script (entry point for Streamlit UI and backend logic)
• Requirements and config:
• requirements.txt – Python dependencies (Streamlit, requests, Gemini client, various APIs)
• .gitignore – standard excludes
• README.md – contains usage instructions, feature list, and directory overview
Extract architectural insights from README
• Streamlit handles the frontend interface
• The app invokes a selection module (powered by Google Gemini) to decide which OSINT tools to run
• It calls multiple CLI tools (theHarvester, amass, etc.) and HTTP-based OSINT APIs (crt.sh, Shodan, AbuseIPDB, etc.)
• Results feed back into the LLM for summarization into a Markdown report
• The UI offers a download link for the generated report
Main components
a. Streamlit Frontend
– UI layer (forms for target input, buttons to install tools, execute recon, download report)
b. Orchestration Layer (Python logic in oh-my-osint.py)
– Gemini Tool-Selector (LLM API client)
– CLI-Tool Runner (wrapper to invoke theHarvester, amass, sublist3r, etc.)
– API Integrator (HTTP clients for crt.sh, ipinfo.io, Shodan, AbuseIPDB)
– Data Aggregator (collects all outputs)
– Gemini Report Generator (passes aggregated data to LLM)
– Report Formatter (constructs Markdown)
c. External Services
– Google Gemini API
– OSINT APIs (crt.sh, ipinfo.io, Shodan, AbuseIPDB, viewdns.info)
d. Local/Downloadable Output
– Markdown report stored in memory or temp storage, served via Streamlit download button
Relationships and data flow
• User → Streamlit UI → Orchestrator
• Orchestrator → Gemini Tool-Selector → returns list of tools
• Orchestrator → CLI-Tool Runner → executes tools, captures STDOUT
• Orchestrator → API Integrator → fetches additional data
• Aggregator collates CLI + API outputs → Gemini Report Generator
• Gemini returns summary → Report Formatter → Markdown
• Streamlit UI ← serves Markdown + Download control
Architectural patterns and principles
• Clear separation of concerns:
• UI vs Orchestration vs External integrations
• Modular design:
• Each tool/integration as a separate wrapper module
• LLM calls encapsulated in a “LLM client” component
• Pluggability/extensibility:
• New CLI tools or APIs can be added via new modules without touching UI code
Essential elements
• Component names and boundaries
• External dependencies clearly separated
• Brief annotation of technology (e.g., “Streamlit (Python)”, “Gemini LLM API”)