Class BlackboardStrategy
- Namespace
- JD.AI.Core.Agents.Orchestration.Strategies
- Assembly
- JD.AI.Core.dll
Blackboard strategy — agents operate on a shared knowledge base (the scratchpad) in iterative rounds. Each round, agents read the current blackboard state, contribute their analysis, and write results back. Converges when no agent contributes new information or max iterations reached. Good for: complex analysis, multi-perspective investigation, knowledge synthesis.
public sealed class BlackboardStrategy : IOrchestrationStrategy
- Inheritance
-
BlackboardStrategy
- Implements
- Inherited Members
Properties
BoardPromptTemplate
Template for the blackboard prompt passed to each agent. Supports {blackboard_state} and {agent_role} placeholders.
public string BoardPromptTemplate { get; init; }
Property Value
MaxIterations
Maximum number of convergence rounds.
public int MaxIterations { get; init; }
Property Value
Name
The strategy name (sequential, fan-out, supervisor, debate).
public string Name { get; }
Property Value
Methods
ExecuteAsync(IReadOnlyList<SubagentConfig>, TeamContext, ISubagentExecutor, AgentSession, Action<SubagentProgress>?, CancellationToken)
Execute the orchestration strategy with the given agents, context, and executor.
public Task<TeamResult> ExecuteAsync(IReadOnlyList<SubagentConfig> agents, TeamContext context, ISubagentExecutor executor, AgentSession parentSession, Action<SubagentProgress>? onProgress = null, CancellationToken ct = default)
Parameters
agentsIReadOnlyList<SubagentConfig>contextTeamContextexecutorISubagentExecutorparentSessionAgentSessiononProgressAction<SubagentProgress>ctCancellationToken