600,000–1 million households displaced. 28+ days of internet blackout. Most humanitarian reports available only in English. CrisisBridge is building a multilingual bridge.
An open-source platform ensuring life-saving humanitarian information reaches every civilian, in their language, online or offline. Currently serving the 2026 Middle East crisis in 12 languages: Arabic, Lebanese Arabic, Persian, Dari, Pashto, Kurdish (Kurmanji), French, Spanish, Russian, English, Chinese, and Turkish.
Read in your language: فارسی | العربية | 中文 | Français | Español | Русский | English
| Stat | Reality | Source |
|---|---|---|
| 🏚️ 600K–1M households | Displaced in Iran alone in the current escalation | UNHCR Flash Refugee Response Plan, March 2026 |
| 📵 28+ days | Iran internet blackout affecting ~90 million people | NetBlocks, March 2026 |
| 👥 33,000 | Afghan refugees in Iran receiving WFP food assistance | WFP Situation Report #4, March 26, 2026 |
| 📄 ~90% | Humanitarian situation reports published in English only | OCHA, ReliefWeb language analysis |
Humanitarian organizations produce critical situation reports daily — predominantly in English. Affected populations in Iran, Afghanistan, Lebanon, and Syria, who primarily speak Arabic, Persian, and Dari, lack access to this information.
CrisisBridge addresses this gap.
We believe in being upfront about where this project stands:
- Telegram is unreachable in Iran. With 28+ days of near-total internet blackout and connectivity at ~1% of normal levels, our primary distribution channel does not reach the population that needs it most. The offline mesh layer (Briar + Meshtastic + D2C satellite) is now a P0 MVP deliverable.
- Persian, Dari, Pashto, Kurdish, and Lebanese Arabic glossary terms are unverified. These are our most critical target languages, but zero terms have been validated by native speakers yet. Translations may contain errors that matter in life-or-death contexts. We urgently need native speaker reviewers.
- Offline mesh transport is implemented but untested in the field.
src/offline/mesh.pyprovides Briar Mailbox and Meshtastic serial integration with compression, fragmentation, and dedup — but no field deployment has occurred yet. - No persistent storage. Reports and user preferences are not persisted. Restarting the system loses all state.
- ACLED API integration pending. The airstrike alert module polls OCHA Flash Updates but the ACLED real-time feed requires an API key not yet configured.
If you can help with any of these, please open an issue or reach out at aid@agentmail.to.
git clone https://github.com/CuiweiG/openclaw-humanitarian.git
cd openclaw-humanitarian
pip install -r requirements.txt
python src/demo.py # See it work in 30 secondsThe demo fetches a real report from the ReliefWeb API and demonstrates the translation pipeline with glossary-based terminology enforcement.
📱 Telegram Bot: @openclaw_aid_bot
Send /start → choose your language → explore available crisis bulletins in your language.
| Product | Status | Description |
|---|---|---|
| 📡 Crisis Scraper | ✅ Live | Auto-scrapes situation reports from ReliefWeb and OCHA APIs |
| 🌐 Multilingual Translator | ✅ Live | AI-powered translation with humanitarian-specific glossary (40 terms, 12 languages, verification tracked) |
| 🤖 Telegram Bot | ✅ Live | Delivers bulletins in 12 languages on demand and via push alerts |
| 🔍 Source Verification | ✅ Live | Credibility scoring for all information sources |
| 🚨 Airstrike Alert System | 🔒 Restricted | Experimental — requires multi-source verification, human review, and partner SOP before activation |
| 📶 Offline Mesh Layer | 🔨 In Dev | Briar + Meshtastic + D2C satellite for internet-denied environments |
| 📱 SMS Gateway | 🔨 In Dev | SMS delivery with HMAC verification; Iran region paused for safety |
| 📦 Supply Chain Tracker | 🔨 In Dev | WFP/OCHA logistics status for humanitarian coordinators |
┌─────────────────────────────────────────────────────┐
│ DATA SOURCES │
│ ReliefWeb API · OCHA HDX · UN Agencies RSS │
└────────────────────────┬────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ SCRAPER PIPELINE │
│ src/scraper/reliefweb.py · src/scraper/ocha.py │
└────────────────────────┬────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────┐
│ TRANSLATION ENGINE │
│ AI model + glossary.json · 40 terms · 12 languages │
│ Quality checks · Source verification │
└────────────────────────┬────────────────────────────┘
│
┌────┴────┬──────────┐
▼ ▼ ▼
┌──────────┐ ┌────────┐ ┌──────────┐
│ Telegram │ │ Alert │ │ Offline │
│ Bot │ │ System │ │SMS/Mesh/ │
│ │ │(OCHA) │ │Satellite │
└──────────┘ └────────┘ └──────────┘
Full architecture details: docs/architecture.md
Here's an illustrative example of a translated bulletin (data shown is for demonstration):
🚨 گزارش بحران — ۲۷ مارس ۲۰۲۶
📍 موقعیت: سوریه شمالی
👥 آوارگان: ۴۲۰٬۰۰۰ نفر
🏥 وضعیت پزشکی: بحرانی — ۳ بیمارستان فعال
🍞 توزیع غذا: روزهای دوشنبه و چهارشنبه، ساعت ۸–۱۲
🚪 گذرگاههای باز: باب الهوا (شمال)
منبع: OCHA، ۲۶ مارس ۲۰۲۶
@openclaw_aid_bot
The project maintains a 40-term humanitarian glossary (with per-term verification tracking) to ensure translation accuracy across all languages:
| English | العربية | فارسی | 中文 |
|---|---|---|---|
| Displaced person | نازح | آواره | 流离失所者 |
| Evacuation route | طريق الإخلاء | مسیر تخلیه | 疏散路线 |
| Emergency shelter | مأوى طارئ | پناهگاه اضطراری | 紧急避难所 |
| Food distribution | توزيع الغذاء | توزیع غذا | 食品分发 |
📖 Full glossary: data/glossary.json
⚠️ Persian (fa), Dari (dar), Pashto (ps), Kurdish (ku), and Lebanese Arabic (apc) translations are machine-generated and unverified. Native speaker review is our top priority. See Contributing and our glossary review workflow if you can help.
✅ Live — deployed and functional | 🔨 In Dev — code exists, not production-ready | 📋 Planned — documented, no code yet
- 🔨 Briar + Meshtastic offline layer — promoted to MVP core
- 🔨 GPS-free routing via Cell Tower ID and governorate lookup tables
- 📋 D2C satellite interface (experimental — Starlink/AST subject to jamming risk in target regions)
- 🔨 SMS gateway with HMAC verification + Iran region pause protocol
- ✅ Source credibility scoring engine (T1–T4 trust tiers)
- 📋 D2C satellite API interface (experimental — subject to GPS jamming and availability constraints)
- 🔨 Pashto and Kurdish (Kurmanji) glossary terms — 40 terms added, unverified
- 🔨 Lebanese Arabic (Levantine) dialect support for 800K+ displaced in Lebanon
- ✅ Multilingual crisis bulletins (AR/APC/FA/DAR/PS/KU/EN/ZH/TR/FR/ES/RU)
- ✅ ReliefWeb/OCHA auto-scraping pipeline
- ✅ 40-term humanitarian glossary (12 languages, per-term verification tracking)
- ✅ Telegram bot with 12-language support
- ✅ Simplified language + emoji markers for low-literacy users
- 🔒 Family links tracing system (restricted — requires DPIA + partner SOP + ICRC coordination)
- 🔒 Mental health & psychosocial support (restricted — self-help content + helpline referral only; no clinical triage)
- 🔨 UXO/mine safety education module
- 🔨 Document protection guide
- 🔨 Security canary system for bot scraping detection
- 🔨 Supply chain status tracker for humanitarian coordinators
- 📋 Audio bulletin generation (pluggable TTS backend)
- 📋 Emergency contact database (IR/LB/AF/SY)
- 📋 Real-time shelter tracker for Lebanon
- 📋 Browser extension for inline report translation
This project builds on peer-reviewed research in humanitarian NLP, crisis informatics, and offline communication systems. We maintain an active research bibliography with 40+ references from IEEE, Frontiers, ICWSM, and leading humanitarian organizations.
Key research areas:
- Humanitarian NLP — Automated extraction of crisis-relevant information from unstructured reports (Rocca et al. 2023, IBM/WFP 2020)
- Low-resource language MT — Machine translation for languages with limited training data (CLEAR Global Gamayun)
- Offline mesh networks — LoRa-based decentralized communication for internet-denied environments (IJARCCE 2025, Meshtastic)
- Community feedback loops — Two-way humanitarian information exchange via chatbots (TWB Uji, IRC Signpost)
- Crisis data standards — HumSet-compatible output format for interoperability with UN systems (OCHA HumSet)
We actively seek collaboration with researchers. If your work intersects with ours, please open an issue or email aid@agentmail.to.
No media coverage yet. If you write about this project, please open a PR to add your link here.
We believe in transparency about what this system can and cannot do:
- AI translation is imperfect. Especially for low-resource languages like Dari, translation quality varies. We maintain a verified glossary and welcome native speaker corrections.
- We are not a replacement for professional humanitarian agencies. Our role is information bridging, not aid delivery.
- LLMs can hallucinate. All AI-generated content goes through our source verification system. Every claim links to its original source.
- Offline mesh networks are experimental. The Briar + Meshtastic layer is a research initiative, not a deployed solution yet.
- User safety is paramount. In countries with internet shutdowns, using communication tools can carry personal risk. We collect only in-memory language preferences (keyed by hashed user ID, lost on restart). See our data flow map for the full audit.
See our humanitarian compliance framework for detailed privacy and ethics policies.
We welcome translators, developers, humanitarian professionals, and volunteers of all backgrounds.
- 🐛 Report a bug
- ✨ Request a feature
- 🌐 Contribute a translation
- 📖 Read the full guide: CONTRIBUTING.md · بالعربية · به فارسی
All skill levels welcome. Native speakers of Persian, Dari, Turkish, Pashto, and Kurdish who can verify translations are especially valued.
This project does not accept financial contributions. To support humanitarian operations directly, please consider donating to established organizations:
- 🔴 ICRC — icrc.org/en/donate
- 🔵 UNHCR — donate.unhcr.org
- ⚕️ MSF / Doctors Without Borders — msf.org/donate
Released under the MIT License. Free to use, modify, and distribute.
- 📧 Email: aid@agentmail.to
- 🤖 Telegram: @openclaw_aid_bot
- 🐛 Issues: GitHub Issues
Developed to bridge the humanitarian information gap. Contributions and visibility help this project reach affected communities.