​
Conflict Detection in OVADARE

Conflict Detection is at the heart of OVADARE’s mission: identifying and addressing issues in multi-agent workflows. It works seamlessly alongside platforms like AutoGen and CrewAI to ensure agents’ actions align with defined policies and do not interfere with each other.

​
What Is Conflict Detection?

Conflict Detection is a framework service that evaluates agents’ actions against predefined policies. It identifies potential issues, such as:

  • Overlapping or conflicting actions between agents.
  • Violations of defined policies or boundaries.
  • Situations requiring intervention to ensure the workflow’s integrity.

​
How Does It Work?

Conflict Detection operates automatically within OVADARE, relying on inputs provided by developers:

  1. Agent Registration
    Agents are registered with OVADARE’s AgentRegistry, allowing the system to track their actions.

    Example:

    Registering an agent with OVADARE
from ovadare.agents.agent_registry import AgentRegistry  
registry = AgentRegistry()  
registry.register_agent(agent_instance)

  2. Policy Definition
    Developers define rules and boundaries for agent actions using the PolicyManager.

    Example:

    Adding a custom policy
from ovadare.policies.policy_manager import PolicyManager
policy_manager = PolicyManager()
policy_manager.add_policy(custom_policy)

  3. Action Monitoring
    OVADARE’s ConflictDetector continuously evaluates agent actions to detect violations.

    Example:

    Detecting conflicts in agent actions
from ovadare.conflicts.conflict_detector import ConflictDetector
detector = ConflictDetector(policy_manager=policy_manager)
conflicts = detector.detect(agent_id="123", action={"type": "write_data"})

  4. Conflict Analysis
    Detected conflicts are logged and made available for further analysis. Developers can review conflicts and decide how to handle them.

​
What Can Developers Do?

Conflict Detection is an automated service, but developers interact with it by:

  1. Defining policies and registering agents to shape the rules governing agent interactions.
  2. Consuming conflict data to improve workflows or resolve issues.

For example:

  • Create policies to prevent resource access conflicts.
  • Use detected conflicts to optimize agent collaboration.

​
Example Workflow

Consider a sales strategy meeting preparation:

  • Agents: A research agent, a strategy agent, and a coordination agent.
  • Task: Gather data, create a strategy, and compile a briefing document.

If two agents attempt to edit the same document simultaneously, OVADARE detects this conflict by:

  1. Evaluating the actions against the policies.
  2. Identifying the overlapping task.
  3. Logging the conflict for resolution.

​
Key Takeaways

  • Conflict Detection is not something developers implement directly but a service layer that monitors workflows.
  • Developers focus on defining agents and policies while OVADARE handles monitoring and evaluation.
  • Outputs include conflict logs and recommendations for resolution.

Ready to define your policies? Check out the Policy Management guide.

QuickstartResolution Engine