| Document | Description |
|---|---|
| CONFIGURATION.md | Configuration reference — docproc.yaml schema, AI providers (OpenAI, Azure, Anthropic, Ollama, LiteLLM), ingest options (vision, LLM refinement). Used by the CLI for document extraction. |
| AZURE_SETUP.md | Azure setup — Azure OpenAI and Azure AI Vision (Computer Vision) for PDF image extraction; credentials via env or scripts/azure_env.sh. |
| ARCHITECTURE.md | Architecture overview — Pipeline flow, modules. docproc is CLI-only (file in → markdown out); the full-stack demo lives in demo/. |
| USAGE.md | Usage examples — CLI extract, init-config, completions. |
| DOCKER.md | Docker — Demo infrastructure only (PostgreSQL, LocalStack, RabbitMQ). Go API and frontend run on the host. |
See also CONTRIBUTING.md for development setup and running tests.
- docproc (CLI) — Document processor only. Reads a file (PDF, DOCX, PPTX, XLSX), extracts text (native + optional vision for images), optionally refines with an LLM, and writes markdown to a file. No server, no database, no RAG.
- Demo — Separate application in
demo/: Go API, React frontend (demo/web/), document upload (LocalStack S3), job queue (RabbitMQ), RAG and grading (PostgreSQL + PgVector, OpenAI). Document processing is done by running the docproc CLI from the Go worker. See demo/README.md.