# ESG Evidence Studio

A retrieval-augmented generation application for searching environmental, social and governance policy documents, inspecting the citations behind an answer, and producing traceable reports, running with no external infrastructure required.

ESG Evidence Studio combines a FastAPI backend with a Streamlit workspace and supports a zero-infrastructure local retrieval mode, backed by an in-memory lexical store, alongside optional Chroma or Qdrant vector-store providers using OpenAI embeddings; the default memory-provider mode needs no Docker container or API key to search and upload documents, though it is intended for development and small demonstrations rather than data that must survive a restart. Document extraction covers PDF, DOCX, TXT, MD, CSV and XLSX files, and a taxonomy of ESG frameworks and categories is kept in a separate configuration file.

A rewrite described as version 3 replaced an earlier, much larger Streamlit prototype with a smaller API-driven frontend, removed fabricated analytics, compliance scores and local report fallbacks that the earlier version had produced, and added typed configuration, constant-time admin-token validation, safe error responses, request IDs, CORS controls, automated tests, linting, CI and non-root containers. Generating an authenticated report requires setting a long, random admin token, and generated answers require an OpenAI or Anthropic API key.

It targets Python 3.11 through 3.14, can also run through Docker Compose, and its license is not asserted in its repository metadata.

Website: https://github.com/mominalix/ESG-Analysis-Using-Retrieval-Augmented-Generation-Engine