RAG Manager Overview

This Symfony-based portal orchestrates how technical teams prepare and oversee Retrieval-Augmented Generation (RAG) content pipelines linked to their GitHub repositories.

Purpose

• Provide a single control plane to connect to a GitHub repo, browse its tree, and flag documents as approved for ingestion.
• Keep governance and visibility on what enters the RAG knowledge base, while a dedicated Python service runs ingestion and vectorization.
• Reduce manual oversight by centralizing status tracking, auditability, and handoff between product, ops, and data teams.
• Provide a chat based interface to ask the RAG system through a local or remote LLM.
• Provide a "Knowledge Graph" based on LightRag's to overview the global knowledge tree.

Scope and Boundaries

• In-scope: GitHub repo linkage, repository tree visualization, document eligibility marking ("ingestable"), and ingestion initiation/monitoring.
• Out-of-scope: The ingestion engine itself (handled by an external Python project: RedwanK/rag-ingestor) and downstream RAG serving or retrieval logic.
• Deployment context: Symfony web app; integrates with GitHub APIs and delegates ingestion calls to the Python backend.

Target Users and Outcomes

• Project/operations leads: ensure only validated content flows into RAG; gain traceability and readiness signals.
• Tech leads/data owners: configure repo access, review eligibility, and trigger ingestion without diving into infrastructure.
• Contributors: see whether their documentation is cleared for RAG and what remains pending.

Value Proposition

• Repository-aware UX: tree view mirrors GitHub structure to avoid mismatches and missed files.
• Separation of concerns: Symfony app handles control and visibility; Python service focuses on ingestion performance.
• Faster onboarding: clear cues on what is “OK” to ingest and what still needs review.

Collaboration Model

1. Link a GitHub repository and sync its structure.
2. Mark eligible documents as “OK” for ingestion within the UI.
3. Dispatch ingestion jobs to the external Python service and monitor completion signals.

Documentations

• Installation : docs/technical/00-installation.md

Functional documentations are available under docs/functional. I'm writing down here my dev planned features and roadmap.