IntelliSearch
IntelliSearch is an MCP-rooted search agent evolved into a decoupled, lightweight infrastructure for the Agentic ecosystem. By unifying context memory, dynamic tool scheduling, and multi-agent communication, it provides developers with a perfectly balanced and adaptable framework.
Everything you need for intelligent search and operations
Introduce IntelliSearch-Toolkit, a high quality toolkit including multi-source search tools and safe operation tools with environment interactions. It also supports extensible MCP-based tools for customization and advanced workflows.
Integrated multi-source search engines including Web (Google/ZHIPU), Scholar, GitHub, Bilibili, WeChat, and more. Supports both Chinese and English content retrieval with intelligent domain selection.
Safely interact with external environments through MCP-based tools. Execute terminal commands, read/write files, operate browsers, and run Python code with SecurityManager protection and fine-grained permission control.
Supports pluggable long-term memory systems for maintaining conversation context across sessions. Sequential memory architecture enables agents to remember and reason over extended interactions with persistent storage.
Provides a pluggable agent architecture supporting multiple agent structures for different tasks. Built on MCP-based framework with extensible design for custom agent development and seamless integration.
Offers both Web and CLI backend modes for flexible deployment. One-click setup with zero configuration for quick start, while supporting deep customization for open-source developers.
Watch how IntelliSearch transforms your workflow
Navigating complex time constraints and multi-modal transit challenges, IntelliSearch leverages multi-source cross-verification to deliver a refined travel plan that balances cost-effectiveness, strict scheduling, and safety redundancy. The solution provides minute-accurate itineraries, cost estimates, and contingency backups, while maintaining full transparency and security by documenting the entire tool-invocation process for verifiable decision-making.
IntelliSearch implements a "Breadth-First, Depth-Follow" tool orchestration paradigm, utilizing a layered strategy of wide-scale recall followed by intensive verification to prevent information gaps and suppress noise accumulation. This framework converges massive data into high-confidence conclusions, delivering structured bibliographies—complete with authorship, direct links, and audit references—to ensure both comprehensiveness and verifiability.
IntelliSearch has significantly upgraded its local data capabilities with a 10,000+ word SAI-exclusive knowledge base, meticulously developed by the Geek Center.
By integrating official media, regulatory documents, and campus archives into a structured hub, it serves as a dedicated information nucleus that empowers SAIers in their daily affairs and academic planning.
Choose your preferred way to experience IntelliSearch - whether you're a student at SJTU or an open-source enthusiast exploring AI-powered search.