Agentic Chip Verification
This challenge focuses on leveraging advanced agentic AI to automate aspects of chip design verification. Participants will build a multi-agent system using LangGraph to orchestrate complex verification workflows, integrating specialized tools via the MCP pattern. The system will employ Gemini 2.5 Pro, specifically its Deep Think mode, for sophisticated reasoning on design specifications and test results, adapting its thought process based on verification complexity. The goal is to demonstrate how agentic AI can streamline highly technical engineering processes by autonomously managing tasks like test case generation, simulation result analysis, and error identification, significantly reducing manual effort and accelerating time-to-market for chip designs. This requires designing a robust, extensible agent architecture capable of both instant decision-making and deep, multi-step analysis.
What you are building
The core problem, expected build, and operating context for this challenge.
Build a graph-based multi-agent system using LangGraph and Gemini 2.5 Pro for automated chip design verification with MCP tool integration.
How work is evaluated
The solution will be evaluated based on the functional correctness of the agentic verification workflow, the effective use of LangGraph for orchestration, the quality of tool integration via MCP, and the demonstration of Gemini 2.5 Pro's advanced reasoning capabilities. Emphasis will be placed on how adaptively the system handles varying verification scenarios.
Challenge datasets
Public datasets, plus the private datasets you have access to.
What you should walk away with
Integrate specialized external tools into an agentic system via the MCP pattern.
Implement adaptive thinking budgets to optimize agent reasoning efficiency.
Design and implement a stateful, graph-based agent workflow using LangGraph for complex engineering tasks.
Utilize Gemini 2.5 Pro's Deep Think mode for advanced, multi-step reasoning.
Reference links and supporting material
Sample data for 1 tasks
Challenge setup
The solution will be evaluated based on the functional correctness of the agentic verification workflow, the effective use of LangGraph for orchestration, the quality of tool integration via MCP, and the demonstration...
Challenge input
{"design_spec": "String (simplified RTL-like description)", "error_type": "String (e.g., 'timing_violation', 'logic_bug')", "tool_stubs": {"simulat...
Agent execution
No agent tool is listed for this challenge.
Evaluated output
{"test_plan": "String", "simulation_summary": "String", "identified_errors": ["String"], "reasoning_path": ["String"]}
- Checks if the LangGraph workflow executes without errors and transitions correctly.
- Verifies if the agent system correctly identifies the simulated error.
- Assesses if the agent successfully invoked the provided tool stubs.
- ReasoningPathLength target: 5
- 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
- Checks if the LangGraph workflow executes without errors and transitions correctly.
- Verifies if the agent system correctly identifies the simulated error.
- Assesses if the agent successfully invoked the provided tool stubs.
Metrics
- ReasoningPathLength · Target 5
- AdaptiveBudgetUsage · Target 0.7
Workflow tasks
- Task 1
Input: {"design_spec": "String (simplified RTL-like description)", "error_type": "String (e.g., 'timing_violation', 'logic_bug')", "tool_stubs": {"simulator_output_function": "python_callable"}}
Output: {"test_plan": "String", "simulation_summary": "String", "identified_errors": ["String"], "reasoning_path": ["String"]}
Execution harness: Python
Recipe freeze marker
No recipe freeze time is recorded.