English | 中文 | 日本語 | 한국어 | Español | हिन्दी | العربية
Open-source collective intelligence engine. 10,000+ AI agents read the open internet daily, form evidence-backed beliefs, debate contested claims, and deliver auditable intelligence across 31 beats.
Live Demo | Documentation | Discord | Report Bug
Watch the 2-minute demo -- see the agent society, belief graph, and intelligence reports in action.
OpenFishh is a persistent collective intelligence platform that deploys thousands of AI agents to read the open internet. Unlike chatbots that answer one question and forget, OpenFishh runs a living society of agents 24/7 -- beliefs compound, sources are re-evaluated, contradictions are debated.
Not a chatbot. Not a simulator. A living intelligence system.
| Feature | Description |
|---|---|
| 10,000+ Agents | Configurable swarm with 7 cognitive roles (scout, researcher, cartographer, infiltrator, tracker, analyst, qualifier) |
| 31 Intelligence Beats | Geopolitics, AI, markets, cybersecurity, healthcare, climate, crypto, defense, and 23 more |
| Epistemic Framework | 5 claim types, 10 source tiers, confidence decomposition, known unknowns, falsification criteria |
| Evidence-Backed | Every belief traces to a source. Every source is scored. Every uncertainty is surfaced |
| Blueprint Reports | Generate auditable intelligence dossiers with trust layers and "What Would Change Our Mind" sections |
| Knowledge Graph | Entity-relationship visualization across all beats with beat-colored clustering |
| Zero API Keys Required | Works with DuckDuckGo search out of the box. Add Brave/Tavily/SearXNG for more coverage |
Configure agents, assign roles across 31 intelligence beats.
Agents read RSS feeds, compress, extract beliefs with epistemic metadata.
Browse the knowledge graph: entities, connections, confidence bands.
Generate auditable intelligence dossiers from accumulated knowledge.
Explore agents, entities, contested beliefs, and epistemic scorecard.
graph LR
A[RSS Feeds<br/>31 Beats] --> B[10K+ Agents<br/>Read & Extract]
B --> C[Beliefs<br/>with Epistemic Metadata]
C --> D[Corroboration<br/>& Debate]
D --> E[Knowledge Graph<br/>& Reports]
- Python 3.12+
- Node.js 18+
- SQLite (included)
# Clone the repository
git clone https://github.com/MohdTalib0/OpenFishh.git
cd OpenFishh
# Backend setup
cd backend
pip install -r requirements.txt
# Frontend setup
cd ../frontend
npm install# Copy environment template
cp .env.example .env
# Required: Set at least one LLM provider
# OpenRouter (recommended, many free models available)
OPENROUTER_API_KEY=your-key-here
# Optional: Search providers (DuckDuckGo works with zero keys)
BRAVE_API_KEY= # 2000 free searches/month
SEARXNG_URL= # Self-hosted, unlimited# Terminal 1: Backend
cd backend
uvicorn app.main:app --reload --port 8000
# Terminal 2: Frontend
cd frontend
npm run devOpen http://localhost:5173 and you're live.
docker compose upFrontend on port 5173, backend on port 8000.
OpenFishh/
├── frontend/ # React + Vite
│ ├── src/
│ │ ├── pages/ # Console (5-step demo), Landing page
│ │ ├── components/ # BeliefGraph (D3), NavBar, ClaimCard
│ │ └── data/demo.json # Real production data (261 entities, 961 beliefs)
│ └── public/ # Fish logo, favicons
│
├── backend/
│ ├── app/
│ │ ├── api/ # FastAPI routes (investigate, society, cycle)
│ │ ├── agents/ # Searcher, Extractor, Epistemics helper
│ │ ├── epistemics/ # Claim types, contradictions, scorecard
│ │ ├── society/ # Daily cycle engine, agent spawning
│ │ ├── report/ # Blueprint report generator with trust layer
│ │ └── feeds.py # 31-beat RSS feed configuration
│ └── scripts/ # spawn_society.py, run_cycle.py
│
├── static/images/ # Logos and icons
├── docker-compose.yml
└── LICENSE # Apache 2.0
What makes OpenFishh different from generic AI tools is the epistemic contract -- every piece of intelligence has metadata about how much you should trust it.
observation -> claim -> hypothesis -> forecast -> recommendation
wire > major_news > specialist_trade > research_preprint > institutional > social > reference > aggregator > unknown
| Band | Confidence | Meaning |
|---|---|---|
| Well-supported | 0.85+ | Multiple independent sources confirm |
| Supported | 0.65-0.84 | Credible sources, moderate corroboration |
| Tentative | 0.45-0.64 | Limited evidence, single source |
| Speculative | <0.45 | Weak evidence, needs investigation |
Every report explicitly states what the system doesn't know. No false confidence.
Click to expand all beats
| Beat | Focus |
|---|---|
| geopolitics | International relations, conflicts, diplomacy |
| ai_startups | AI companies, funding, product launches |
| ai_research | Papers, models, benchmarks, breakthroughs |
| markets | Stock markets, commodities, macro indicators |
| cybersecurity | CVEs, APTs, threat actors, incidents |
| healthcare | Public health, FDA, WHO, pharma |
| climate_energy | Renewables, fossil fuels, climate policy |
| economics | Central banks, inflation, trade, employment |
| crypto_web3 | Bitcoin, Ethereum, DeFi, regulation |
| defense_govt | Military, defense spending, intelligence |
| regulation | AI policy, antitrust, data privacy |
| biotech_pharma | Drug development, clinical trials, CRISPR |
| supply_chain | Semiconductors, shipping, rare earths |
| social_trends | Remote work, mental health, Gen Z |
| media_entertainment | Streaming, gaming, content industry |
| dev_tools | IDEs, frameworks, open source tools |
| vc_funding | Venture capital, seed rounds, exits |
| frontier_tech | Quantum, robotics, space, neurotech |
| consumer_retail | E-commerce, retail trends, consumer spending |
| education | EdTech, online learning, policy |
| culture_philosophy | Ethics, philosophy, cultural movements |
| real_estate | Housing markets, commercial RE |
| food_agriculture | AgTech, food security, supply |
| global_south | Emerging markets, development |
| sports | Sports business, analytics |
| science_space | Space exploration, physics, astronomy |
| saas_market | SaaS trends, PLG, enterprise software |
| competitive_intel | M&A, market positioning |
| india_startups | India tech ecosystem |
| india_edtech | India education technology |
| general_tech | Broad technology news |
| OpenFishh | ChatGPT / Perplexity | MiroFish | |
|---|---|---|---|
| Approach | Persistent multi-agent society | Single-query chatbot | Closed-world simulation |
| Data source | Open internet (RSS, news, research) | Training data + web search | User-uploaded documents |
| Persistence | Beliefs compound over time | No memory between queries | Per-simulation only |
| Auditability | Every claim has source, tier, confidence | "Trust me" | Report-level |
| Scale | 10,000+ agents, 31 beats | 1 model | Hundreds of agents |
| Cost | Free (DuckDuckGo + free LLMs) | $20-200/month | Requires API keys |
| Open source | Yes (Apache 2.0) | No | Yes (Apache 2.0) |
# Spawn 500 agents across 15 beats
python backend/scripts/spawn_society.py --agents 500 --beats 15
# Run a daily cycle
python backend/scripts/run_cycle.py
# View the scorecard
curl http://localhost:8000/api/scorecard| Method | Endpoint | Description |
|---|---|---|
| POST | /api/spawn |
Spawn a new society |
| POST | /api/cycle/run |
Run daily cycle (SSE streaming) |
| GET | /api/stats |
Society statistics |
| GET | /api/beliefs |
Browse all beliefs |
| GET | /api/beliefs/contested |
Contested beliefs with opposing stances |
| GET | /api/beings |
List active agents |
| GET | /api/entities |
Entity list with mention counts |
| POST | /api/investigate |
Generate Blueprint report (SSE) |
| GET | /api/report/:id |
Retrieve a generated report |
| GET | /api/scorecard |
Epistemic health scorecard |
These numbers are from our running production society:
| Metric | Value |
|---|---|
| Active agents | 1,200 |
| Total beliefs | 37,563 |
| Entities tracked | 16,824 |
| Intelligence beats | 31 |
| Forecast accuracy | 85.7% (6/7 verifiable) |
We welcome contributions! See our issues page for open tasks.
# Fork, clone, and create a branch
git checkout -b feature/your-feature
# Make changes, test, and submit a PRApache 2.0. See LICENSE for details.
OpenFishh is built by @MohdTalib0. The epistemic framework, society engine, and intelligence pipeline are informed by research in collective intelligence, epistemic logic, and multi-agent systems.
openfishh.com | GitHub | Docs
If OpenFishh helps your research or work, please consider giving it a star.