Frontend Usage¶
This page covers the concrete frontend flow for standard and LLM full-text extraction.
Accessing the Frontend¶
- Development:
http://localhost:5734 - API for the frontend:
http://localhost:8734
Running an LLM Full-Text Analysis¶
- Start the API with your LLM environment configured.
- Open the frontend at
http://localhost:5734. - Paste research phenotype text into the main input.
- Open Advanced Options.
- Select the
llmextraction backend. - The shared default model is
gemini-3.1-flash-lite-preview; override it only for targeted experiments. - Keep
two_phaseas the mode. - Submit the analysis.
The results view keeps the same overall structure as standard extraction and adds LLM metadata so you can verify which model and mode produced the output.
Example API Environment for Frontend Validation¶
export PHENTRIEVE_ENV=production
export PHENTRIEVE_TRUSTED_PROXY_CIDRS="127.0.0.1/32,10.0.0.0/8"
export PHENTRIEVE_LLM_DAILY_LIMIT=3
export PHENTRIEVE_LLM_QUOTA_DB_PATH="../data/app/llm_quota.db"
Use those settings when you want the frontend to exercise the same proxy and quota behavior as the production API. The benchmark CLI does not use this quota path.