Open-source Deep Research

Deep Research for LlamaIndex Agents

Let LlamaIndex agents gather cited outside evidence before they plan, retrieve, or synthesize.

Agent tools beyond local indexes

LlamaIndex agents can reason over indexed knowledge and tools. AutoSearch adds public research coverage when the agent needs recent examples, community reports, papers, or Chinese-language sources that are not already inside local indexes.

MCP boundary keeps systems modular

AutoSearch can be attached as an MCP-native capability instead of being folded into each LlamaIndex pipeline. That keeps experimental public research separate from private retrieval, document ingestion, and production index maintenance.

Cited evidence for agent outputs

When agents produce recommendations, users need to inspect the basis for claims. AutoSearch returns cited findings across 40 channels, making it easier for LlamaIndex agents to produce traceable reports and defensible next actions.

How it fits

AutoSearch sits next to LlamaIndex agents as a tool for current external discovery. LlamaIndex continues to manage indexes, query engines, workflows, and synthesis. AutoSearch handles open-source deep research across public channels and returns cited source material. Agents can then combine private retrieval with fresh outside evidence, especially for product research, technical planning, academic monitoring, and multilingual market understanding.

Try this prompt

Find recent examples of LlamaIndex agents using MCP or external tools.
Summarize patterns, caveats, and citations.