AutoGen Multi-Agent UK Planning Portal Zoning Document Reviewer
Deploy a collaborative AutoGen multi-agent system to automate UK Planning Portal document verification and local development plan checking. The multi-agent workflow converses to resolve zoning discrepancies, verifying structural filings and outputting fully structured policy compliance evaluations.
What you are building
Uses Microsoft AutoGen multi-agent conversational patterns to orchestrate automated verification of UK planning applications against municipal frameworks.
How work is evaluated
Evaluates multi-agent conversation resolution and statutory policy check precision.
Datasets
Public datasets, plus the private datasets you have access to.
How submissions are scored
Policy Breach Detection
Verifies that conservation area height limits are correctly identified
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Agent Conversation Efficiency
Number of multi-agent turns required to achieve correct consensus • target: 4 • range: 2-10
This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.
Learning goals
Configure AutoGen multi-agent group chats with specialized planner roles
Automate cross-validation between UK planning applications and local policy documents
Implement human-in-the-loop triggers when policy ambiguity is detected
Generate formal approval recommendation notices with statutory clause citations
Resources
40 UK planning permission applications paired with local authority development framework constraints.
Challenge setup
Evaluates multi-agent conversation resolution and statutory policy check precision.
Challenge input
JSON containing application_id, proposed_extension_sqm, conservation_area (bool), and roofline_height_m
AutoGen
Orchestrates multi-agent conversation and consensus building.
Evaluated output
JSON containing compliance_status, flagged_policy_breaches, and agent_consensus_summary
- Verifies that conservation area height limits are correctly identified
- Number of multi-agent turns required to achieve correct consensus • target: 4 • range: 2-10
- Benchmark: MultiAgentPolicyCheck-UK
- Agent Conversation Efficiency target: 4
- 1 public reference case
- Python execution harness
- Python sandbox (unavailable on Versalist)
Shows the first task, up to three tools, and up to three evidence items.
View technical recipe
Configured tools
- AutoGen · Required
- Larridin · Optional
- Hamming · Optional
Evaluation contract
- Policy Breach Detection · Weight 1
- Agent Conversation Efficiency · Weight 1
Metrics
- Agent Conversation Efficiency · Target 4
Benchmarks
- MultiAgentPolicyCheck-UK
- general
Reference cases
- 1 public reference case
Workflow tasks
- Task 1
Input: JSON containing application_id, proposed_extension_sqm, conservation_area (bool), and roofline_height_m
Output: JSON containing compliance_status, flagged_policy_breaches, and agent_consensus_summary
Execution harness: Python
Evaluation tools
- AutoGen · Multi-Agent Framework
Recipe freeze marker
No recipe freeze time is recorded.