fdarkaou/open-deep-research
Open Deep Research
Open-deep-research is an autonomous research orchestrator and language model-based agent designed to execute multi-step investigations and automated web research workflows. It explores complex topics systematically by generating targeted search queries, processing multiple sources simultaneously, and recursively feeding synthesized findings back into search planning loops to expand its research scope.
The system coordinates these tasks through a graph-based agent orchestration model that routes state between modular nodes for searching, reading, and reasoning. It accelerates data collection via a concurrent web scraping pipeline and compiles aggregated findings into cohesive, cited markdown documents complete with hierarchical headings and reference lists.
During execution, the application streams real-time feedback and incremental report updates to the client interface using server-sent events. Security features include credential isolation that stores user-provided API keys in HTTP-only cookies to protect sensitive access keys from client-side scripts.
Features
- Deep Research Execution - Explores complex topics systematically by generating targeted search queries, processing multiple sources simultaneously, and recursively gathering comprehensive information.
- Graph-Based State Orchestrations - Orchestrates multi-step agent workflows by routing state transitions between modular search, read, and reasoning nodes using directed graphs.
- Automated Research Generation - Explores complex topics systematically by generating targeted search queries and gathering comprehensive information from multiple sources.
- Autonomous Research Frameworks - Coordinates multi-step investigations by generating targeted search queries and recursively gathering information.
- Autonomous Web Research Agents - Explores complex topics systematically by generating targeted search queries and gathering comprehensive information from multiple online sources.
- Cited Report Compilers - Synthesizes aggregated web research into cohesive markdown documents complete with hierarchical headings and reference lists.
- Structured Research Reports - Compiles detailed documents featuring organized findings, referenced sources, and progress tracking for easy viewing and downloading.
- Professional Research Reports - Compiles detailed documents featuring organized findings, referenced sources, and ongoing progress tracking for easy viewing and download.
- Search Query Recursion - Expands research scope dynamically by feeding synthesized findings back into search planning loops to generate targeted follow-up queries.
- Web Research Agents - Executes iterative web searches, processes multiple sources, and compiles comprehensive reports on complex topics.
- Concurrent Scraping Workers - Accelerates data gathering by executing multiple parallel HTTP requests and web parsing tasks across discovered source URLs.
- Self-Hosted AI Infrastructure - Runs open-source deep research workflows locally to emulate advanced AI reasoning and search capabilities for complex investigations.
- Server-Sent Events - Streams real-time execution feedback and incremental report updates to the client interface over persistent HTTP connections.
- Research Progress Dashboards - Displays live visual feedback and incremental findings while the system performs parallel searches and builds the final report.
- Real-time Pipeline Progress Visualization - Displays real-time visual feedback and incremental findings while the system performs parallel searches and builds the final report.
Star history
How this analysis was created: This summary and feature list were written by an AI model that read the project's README and public documentation pages. Each feature links to the documentation it came from; stars, license and language come straight from the GitHub API. The model does not read the source code, and the analysis is refreshed when the project is re-analysed. Learn more on our About page.
