agentic-retrieval

RAGFlow is an agent-native database: a Retrieval-Augmented Generation (RAG) engine that grew into context infrastructure.

DR Pixel: Page-Native Deep Research with Vision-Language Models — benchmark, evaluation harness and training code

Reusable accelerator for Foundry IQ live grounding with Fabric Ontology and MCP Server Knowledge Sources.

An extensive and commented list of resources on agentic search and agentic deep research

Enterprise knowledge infrastructure for RAG and agents. Every answer is traceable to immutable evidence and checked against the live version before it is ever returned.

Agentic code retrieval benchmark for coding agents

AI support agent using Azure Foundry, RAG, Terraform

Reasoning-based, vectorless RAG over a large document using a hierarchical tree (PageIndex) and a Vision-Language Model (Llama 4 Scout), no embeddings, no vector store, no text chunking.

[EMNLP 2026]* SSE-Bio: A Structured Self-Evolving Agent with Agentic Retrieval Policy for Multi-Hop Biomedical Reasoning.

MCP server for Azure AI Search, supporting keyword, semantic, vector, hybrid, and Knowledge Base agentic retrieval.

Local document search with an agentic LLM loop (SQLite FTS5, no embeddings). Answers cite the page. MCP server, Windows & Linux desktop app.

LLM-readable debugging knowledge base: structured pits that help coding agents find buried fixes in human-oriented threads.

A second brain that fills itself: one click on a YouTube video, or a like on LinkedIn or X, and Claude turns it into linked, fact-checked Obsidian notes you can search and ask.

Blackwall is an autonomous Agentic Firewall designed to secure AI agents by actively intercepting execution flows before they reach external systems or the host OS. It leverages self-learning threat signatures, structural graph analysis, and real-time semantic gating to dynamically detect and mitigate rogue agent behaviors and malicious payloads.

A TypeScript demo of Azure AI Search agentic retrieval with NASA data

Data-agnostic Agentic RAG with hybrid BM25, Chroma, and Neo4j retrieval, source verification, bounded retries, and AWS ECS deployment.

Intercepts an AI agent's action before it runs, grounds it in cited precedent via Foundry IQ, pauses for a human.

An enterprise view into Agentic Retrieval for Multi-language usecases