agent-streaming

Stream orchestration events from agents via astream() for real-time progress.

19|6|Updated Mar 12, 2026
One-click install
npx skills add https://github.com/NicolaiLassen/orxhestra --skill agent-streaming
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-streaming
Source: https://github.com/NicolaiLassen/orxhestra/tree/main/docs/skills/agent-streaming
Command: npx skills add https://github.com/NicolaiLassen/orxhestra --skill agent-streaming

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill solves the problem of getting responsive, real-time feedback from long-running multi-agent tasks instead of waiting for a final result.

Core Features & Use Cases

  • Token-by-token streaming: Render partial agent responses as they are produced via astream() and Event objects.
  • Real-time sub-agent event propagation: Stream child agent events through AgentTool, including branch and agent_name for live coordination visibility.
  • Runner-compatible streaming: Consume streamed events from the runner.astream(...) loop with final-response detection.

Quick Start

Use the agent-streaming Skill to stream partial token output while a parent agent delegates work to sub-agents for a live planning result in the terminal.

Frequently Asked Questions about agent-streaming

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I stream partial LLM responses token-by-token in a multi-agent CLI?

Token-by-token streaming is handled by iterating orchestration events asynchronously via `astream()` and rendering `Event` objects containing partial agent messages as they are produced.

How does sub-agent event propagation work during nested agent delegation?

Sub-agent event propagation streams child agent events through `AgentTool` using `branch` and `agent_name` fields, allowing live coordination visibility and branch-aware event rendering for nested tasks.

Can I use asyncio event handling to get real-time feedback from long-running LLM orchestration?

Yes, asyncio event handling supports real-time visibility into LLM orchestration progress by consuming streamed orchestration events from orxhestra agents, preventing users from waiting for a final result.

What is the best way to detect final responses in a Runner-based streaming session?

Runner-based streaming detects final responses by consuming streamed events from the `runner.astream(...)` loop, enabling real-time session updates and final-response detection within the async iteration.

Do I need asyncio to handle multi-agent streaming events?

Yes, multi-agent streaming requires async event iteration via `astream()` to handle EventType agent messages with partial output and to propagate child events through AgentTool correctly.