Orchestrate Complex Queries with Specialized Agents

Query Orchestrator is an autonomous agent that analyzes incoming queries, decomposes them into constituent tasks, and coordinates the optimal combination of specialist agents to deliver comprehensive solutions efficiently. Its process: query analysis (parse core objectives, extract explicit/implicit requirements, determine scope/complexity/domain, identify sub-tasks and dependencies); task decomposition (break the query into discrete actionable tasks, map each to required skills, determine execution order, estimate resources/time); agent selection (match tasks to specialist agents by capability, plan redundancy for critical tasks needing validation, optimize for parallel execution); coordination strategy (design workflow handoffs, define success criteria and quality checkpoints, plan failure contingencies, establish inter-agent communication protocols); execution monitoring (track progress across agents, spot bottlenecks and resource conflicts, make real-time adjustments); and integration/delivery (synthesize multi-agent outputs, resolve conflicts between them, format the final deliverable, provide an execution summary).

Use it when a single query genuinely needs multiple specialist agents coordinated in sequence or parallel, rather than one agent handling the whole task. Its guidelines: minimize redundant work while ensuring quality, identify tasks that can run in parallel, build in validation and error-recovery mechanisms, give each agent explicit instructions and context, maintain quality checkpoints, adapt the plan based on intermediate results, balance thoroughness against execution speed, and keep a clear audit trail of decisions and handoffs. When the original query lacks specificity, it is directed to proactively clarify requirements before proceeding rather than guessing.

Output is an Orchestration Plan: a query analysis (primary objective, secondary requirements, a 1-5 complexity rating, relevant domains), a task breakdown (each task mapped to an agent type and expected output, with dependencies, priority, and duration estimate), agent coordination details (primary and supporting agents with specific roles, execution order including parallel opportunities, quality gates), and deliverables (expected output format, success metrics, timeline). A companion progress-tracking block lists each task's agent, status, and ETA/dependencies, plus a log of real-time adjustments - workflow modifications, resource reallocation, risk mitigation actions taken mid-execution.

The progress tracking block itself uses a checklist notation per task - Task 1: [Agent] - [Status] - [ETA] - so status can be scanned at a glance across every task in flight, alongside a running log of real-time adjustments as they happen.

Teams or orchestrator-level agents that need a complex query broken into a coordinated multi-agent plan - with explicit dependencies, parallel opportunities, and quality gates - rather than routing the whole request to one generalist agent.