How mr-chetan-66/viralens_ai works
ViraLens AI accepts typed text or an image, extracts and classifies factual claims, searches sources, evaluates evidence, computes a verdict and confidence, and presents a plain-language report with citations. The backend orchestration and major verification stages are represented alongside the React interface and OCR. The excerpts show the pipeline’s internal calls; API-to-UI wiring and some search-provider dispatch details were not shown, so those connections are omitted.
Main entry points: utils.py Pipeline orchestration · llm.py LLM client configuration · App.jsx Fact-check UI · api.py Backend API
Components 14
User experience
- Fact-check UI
frontend/src/App.jsx - Backend API
api.py
Claim intake
- Pipeline orchestration
utils.py - Image OCR
ocr.py - Claim extraction
claim_extractor.py
Evidence verification
- Source search and ranking
search_engine.py - Evidence evaluation
evaluator.py
Verdict and reporting
- Verdict and confidence
verdict.py - Plain-language report
reporter.py
Shared services
- Cancellation checks
cancellation.py - LLM client configuration
llm.py
Other components
- User
- Groq inference
- Web search services
Connections 16
- User → Fact-check UI: submits claim
README.md - Fact-check UI → Backend API: requests checks
- Backend API → Pipeline orchestration: delegates processing
api.py - Pipeline orchestration → Cancellation checks: checks cancellation
utils.py - Pipeline orchestration → Claim extraction: classifies claims
utils.py - Pipeline orchestration → Image OCR: extracts image text
utils.py - Pipeline orchestration → Source search and ranking: searches sources
utils.py - Pipeline orchestration → Evidence evaluation: evaluates evidence
utils.py - Pipeline orchestration → Verdict and confidence: computes verdict
utils.py - Pipeline orchestration → Plain-language report: builds report
utils.py - Claim extraction → LLM client configuration: gets client
claim_extractor.py - Evidence evaluation → LLM client configuration: gets client
evaluator.py - Plain-language report → LLM client configuration: gets client
reporter.py - Image OCR → Groq inference: transcribes image
ocr.py - Source search and ranking → Web search services: queries providers
- LLM client configuration → Groq inference: configures client