---
title: OSINT Threat Analyst
emoji: 🌐
colorFrom: indigo
colorTo: blue
sdk: gradio
sdk_version: 5.23.0
app_file: app.py
pinned: false
license: apache-2.0
tags:
  - osint
  - security
  - geopolitics
  - threat-intelligence
  - agentic
  - smolagents
  - conflict-analysis
  - nlp
---

# 🌐 OSINT Threat Analyst

An agentic multi-source open-source intelligence (OSINT) briefing tool that autonomously gathers, fuses, and synthesizes geopolitical conflict data into structured threat assessments.

## Architecture

This Space demonstrates a **context-aware agentic analyst** loop:

1. **Tool use** — The agent autonomously decides when to call ACLED (structured conflict event data) and RSS news feeds (unstructured open-source reporting)
2. **Multi-source fusion** — Raw data from heterogeneous sources is normalized and co-analyzed
3. **Structured synthesis** — Output is produced as a typed threat brief with severity, confidence, key findings, escalation indicators, and watch items
4. **Dual output** — Rendered HTML brief + transparent agent trace for auditability

## Data Sources

| Source | Type | Coverage |
|--------|------|----------|
| [ACLED](https://acleddata.com) | Structured conflict events | 80+ countries |
| Reuters, BBC, Al Jazeera | RSS news | Global |
| Bellingcat, Crisis Group | OSINT/analysis RSS | Conflict-specific |
| UN News, Foreign Policy | Policy/humanitarian | Global |

## Setup (Self-hosting)

Set the following Space secrets:

```
ACLED_API_KEY   — Register free at https://developer.acleddata.com
ACLED_EMAIL     — Email used for ACLED registration
HF_TOKEN        — Auto-injected in HuggingFace Spaces
```

## Research Context

This tool operationalizes the **context-aware agentic security analyst architecture** explored in ongoing research on converged physical/cyber security intelligence operations. The agentic loop pattern — autonomous tool selection, multi-source ingestion, and structured synthesis — mirrors the design described in the supporting academic work.

Key research contributions demonstrated here:
- Local-language and multilingual source integration (extend `RSS_FEED_REGISTRY` in `tools.py`)
- Structured output schema enabling downstream automation
- Transparent reasoning trace for analyst oversight

## Extending

To add new RSS sources, add entries to `RSS_FEED_REGISTRY` in `tools.py`:

```python
RSS_FEED_REGISTRY["your_source"] = "https://example.com/feed.xml"
```

To add new tools (e.g. GDELT, Twitter/X, Telegram), decorate functions with `@tool` and pass them to `ToolCallingAgent`.

## Stack

- [`smolagents`](https://github.com/huggingface/smolagents) — HuggingFace-native agentic framework
- [`Qwen/Qwen2.5-72B-Instruct`](https://huggingface.co/Qwen/Qwen2.5-72B-Instruct) — Free HF Inference API model
- [ACLED API](https://acleddata.com/acleddatanew/wp-content/uploads/dlm_uploads/2021/11/ACLED_API-User-Guide_2021.pdf) — Armed Conflict Location & Event Data
- [Gradio](https://gradio.app) — UI framework

## Disclaimer

⚠️ AI-generated from open sources. Not for operational use without independent verification. ACLED data is third-party and subject to their terms of use.