Build a Hybrid Reasoning Web Research Agent
This challenge involves developing an autonomous agent system capable of comprehensive web research. The system will feature a primary orchestrator agent that intelligently delegates tasks to specialized sub-agents. These agents will leverage advanced Model Context Protocol -enabled web browsing tools and implement a sophisticated Retrieval Augmented Generation (RAG) pipeline to efficiently synthesize information from diverse web sources. A core aspect of this challenge is implementing a hybrid reasoning mechanism: agents will dynamically shift between 'instant' (quick, high-level summarization) and 'deep' (detailed analysis, cross-referencing, critical evaluation) modes using adaptive thinking budgets based on the complexity and novelty of the research task. This approach ensures optimal resource allocation and deeper insights when required, mimicking human-like cognitive processes in an agentic framework.
What you are building
The core problem, expected build, and operating context for this challenge.
Develop a Langroid-based multi-agent system with Claude Opus 4.1 that performs advanced web research using Model Context Protocol-enabled tools and adaptive hybrid reasoning.
How work is evaluated
The evaluation assesses the agent system's ability to conduct comprehensive, accurate, and ethically sourced web research, synthesize findings, and demonstrate effective adaptive hybrid reasoning.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Design and implement a multi-agent system using Langroid for autonomous web research.
Develop a hybrid reasoning mechanism using Claude Opus 4.1 for adaptive problem-solving.
Implement an advanced RAG pipeline for synthesizing web-sourced information efficiently.
Integrate Model Context Protocol (Model Context Protocol)-enabled web browsing and data extraction tools.
Reference links and supporting material
Sample data for 2 tasks
Challenge setup
The evaluation assesses the agent system's ability to conduct comprehensive, accurate, and ethically sourced web research, synthesize findings, and demonstrate effective adaptive hybrid reasoning.
Challenge input
{ "query": "String", "expected_depth": "instant|deep" }
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{ "summary": "String", "sources": [ { "url": "String", "snippet": "String" } ], "reasoning_log": [ { "timestamp": "String", "mode": "instant|deep",...
- Confirms successful invocation and response handling from Model Context Protocol-enabled web browsing tools.
- Ensures the generated summary does not contain factually incorrect information not supported by sources.
- Verifies that all facts in the summary are traceable to provided sources.
- InformationCompleteness target: 90
- Python execution harness
Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.
View technical recipe
Configured tools
No tool records are attached.
Evaluation contract
- Confirms successful invocation and response handling from Model Context Protocol-enabled web browsing tools.
- Ensures the generated summary does not contain factually incorrect information not supported by sources.
- Verifies that all facts in the summary are traceable to provided sources.
Metrics
- InformationCompleteness · Target 90
- FactualAccuracy · Target 95
- ReasoningDepthScore · Target 4
- TimeEfficiency · Target low
Workflow tasks
- Task 1
Input: { "query": "String", "expected_depth": "instant|deep" }
Output: { "summary": "String", "sources": [ { "url": "String", "snippet": "String" } ], "reasoning_log": [ { "timestamp": "String", "mode": "instant|deep", "decision_reason": "String" } ] }
Execution harness: Python
- Task 2
Input: { "query": "String", "expected_mode": "instant|deep" }
Output: { "actual_mode_sequence": ["instant", "deep", ...], "justification": "String" }
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.