manager-getting-started

What problem does it solve?

Manager centralizes multi-AI orchestration so teams do not have to manually choose between backends, stitch parallel runs together, or build brittle workflow chains when delegating coding, review, and reasoning tasks to multiple AI providers.

Core Features & Use Cases

• Task lifecycle & polling: Submit tasks, check status, and retrieve outputs across heterogeneous backends.
• Smart routing & templates: Use auto_route or explicit routing to let Manager learn which backend performs best and save reusable workflow templates.
• Parallel execution & workflows: Run concurrent review or generation steps, chain multi-step workflows with retries and escalation, and group parallel tasks by dependency.
• Interactive sessions & direct calls: Start session-based interactions or call backends directly (Claude Code, Codex, Gemini CLI, GPT) for targeted jobs.
• Analytics & monitoring: Collect execution stats, run analyzers to promote/demote backends, and surface active breadcrumbs in the dashboard.

Quick Start

Ask Manager to submit the coding task Add error handling to src/main.rs with auto_route enabled so it selects the best backend and returns the result.