Add VORTEXRAG — 7-Layer RAG with Semantic Drift + Context Poisoning elimination

[vignesh2027]

VORTEXRAG — Open Source RAG Framework

VORTEXRAG is a complete 7-layer RAG framework that solves two fundamental problems:

1. Semantic Drift — retrieves causally-relevant chunks, not just topically similar ones
2. Context Window Poisoning — provably-optimal purging of low-signal chunks

Key Results

• EM: 74.8 (+13.6 vs Naive RAG, +6.4 vs Self-RAG)
• F1: 82.6 | Faithfulness: 0.94
• Latency: 185ms (2.2× faster than Self-RAG)
• 229 unit tests, MIT license, Python 3.10–3.13

Links

• Code: https://github.com/vignesh2027/VORTEXRAG
• Paper: https://doi.org/10.5281/zenodo.20579702
• Demo: https://huggingface.co/spaces/vigneshwar234/VORTEXRAG

Would love to be added to the RAG or LLM apps section!

[harshil-1402]

Could this be assigned to me? I'd like to add VORTEXRAG to the RAG section — the 7-layer pipeline approach (causal-relevance retrieval + provably-optimal context purging) is a useful addition alongside the existing entries.

I'll open a PR shortly with the entry added in the right place, following the repo's existing format.

[vignesh2027]

Hey @harshil-1402 — yes, please go ahead! Would love to have VORTEXRAG in the RAG section.

A few details that might help you write the entry:

What makes it different from the other RAG entries:

• 7-layer pipeline (not just retrieve + generate) — each layer has a specific job
• Layer 3 (SDC) eliminates semantic drift before chunks enter the context window
• Layer 4 (CPG) purges poisoned/contradictory chunks
• Layer 7 (FV) verifies faithfulness after generation and regenerates if needed
• Faithfulness score: 0.94 vs 0.81 (Self-RAG) and 0.71 (Naive RAG)

Links for the entry:

• GitHub: https://github.com/vignesh2027/VORTEXRAG
• Paper: https://doi.org/10.5281/zenodo.20579702
• Demo: https://huggingface.co/spaces/vigneshwar234/VORTEXRAG

Feel free to ping me here if you need anything else for the PR. Looking forward to it!

[harshil-1402]

Sure thanks a lot for the info, i'll start right away!

[Shubhamsaboo]

Thanks for sharing VORTEXRAG, and credit for the open-source release and test coverage. The fit here is the issue: awesome-llm-apps collects self-contained, runnable tutorial apps, not standalone frameworks/libraries, and adding VORTEXRAG would be a README link to an external repo. The published benchmark numbers are also self-reported and not independently verified, so we'd rather not feature them in the list.

Note for anyone planning a PR from this thread: the earlier "go ahead" reply was from the project's own author, not a maintainer, so please don't open a README-link PR expecting it to be merged. Closing as out of scope. Appreciate the work regardless.

---

Generated by Claude Code

[harshil-1402]

may i know the reason for the issue to be closed as "not planned"