feat: Multi-Agent Orchestration System by emoubarak · Pull Request #2036 · agentscope-ai/QwenPaw

emoubarak · 2026-03-22T01:58:33Z

Summary

This PR implements a complete multi-agent orchestration system that allows agents to dynamically spawn and coordinate with other agents. It includes:

Backend: Orchestration configuration API, workflow engine, and spawn_agent tool
Frontend: UI for configuring orchestration settings and provider/model selection in Agent Management
Tests: Comprehensive unit tests for the spawn tool and workflow engine

Features

Backend

Feature	Description
`spawn_agent` Tool	Async tool allowing agents to spawn other agents with permission checks
OrchestrationConfig	Pydantic model with `can_spawn_agents`, `allowed_agents`, `max_spawn_depth`
API Endpoint	`GET/PUT /api/agent/orchestration-config` for runtime configuration
Context Tracking	`ContextVar` pattern for spawn depth tracking
Permission Checks	Self-spawning prevention, allowed list validation, depth limits

Frontend

Feature	Description
OrchestrationCard	New component in Agent Config page for orchestration settings
Agent Management Modal	Orchestration fields + provider/model cascade selection
Provider Filtering	Show only eligible providers (configured with API key and models)
i18n	Full English and Chinese translations

Architecture

┌─────────────────────────────────────────────────────────────┐
│                      Agent (caller)                          │
│  ┌─────────────────────────────────────────────────────┐    │
│  │  spawn_agent(agent_id, prompt, timeout)            │    │
│  │    ↓                                                │    │
│  │  Permission Checks:                                 │    │
│  │    - can_spawn_agents?                              │    │
│  │    - agent_id in allowed_agents?                    │    │
│  │    - spawn_depth < max_spawn_depth?                 │    │
│  │    - agent_id != current_agent_id? (no self-spawn)  │    │
│  │    ↓                                                │    │
│  │  MultiAgentManager.get_agent(agent_id)              │    │
│  │    ↓                                                │    │
│  │  workspace.runner.query_handler(prompt)             │    │
│  │    ↓                                                │    │
│  │  [Agent: agent_id] Response                         │    │
│  └─────────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────────┘

Files Changed

Backend (Python):
  src/copaw/agents/tools/spawn_agent.py          # Core spawn tool
  src/copaw/agents/react_agent.py               # Tool registration
  src/copaw/app/agent_context.py                # ContextVar for spawn depth
  src/copaw/app/orchestration/                  # Workflow engine + schema
  src/copaw/app/routers/agent.py                # Orchestration config API
  src/copaw/config/config.py                    # OrchestrationConfig model

Frontend (TypeScript):
  console/src/pages/Agent/Config/               # OrchestrationCard component
  console/src/pages/Settings/Agents/            # Agent modal with orchestration

Tests:
  tests/unit/agents/tools/test_spawn_agent.py   # 24 unit tests
  tests/unit/app/orchestration/                  # Engine + schema tests

Test Coverage

Component	Tests	Description
`spawn_agent` tool	24	Input validation, permissions, self-spawning, timeout, response handling
Workflow engine	11	Permission checks, depth validation

Usage Example

# Agent configuration (YAML or API)
orchestration:
  can_spawn_agents: true
  allowed_agents: ["researcher", "writer"]
  max_spawn_depth: 3

# In agent's tool call
spawn_agent(
    agent_id="researcher",
    prompt="Find information about quantum computing",
    timeout=120
)

API

# Get orchestration config
GET /api/agent/orchestration-config

# Update orchestration config
PUT /api/agent/orchestration-config
{
  "can_spawn_agents": true,
  "allowed_agents": ["researcher"],
  "max_spawn_depth": 3
}

Note for Reviewers

If anything needs to be changed, or if you think a different approach would be better, please don't hesitate to leave a comment with your feedback. I'm open to refactoring, adjusting the architecture, or taking an entirely different direction if that makes more sense for the project. Your feedback is very welcome! 🙏

Implement a complete multi-agent orchestration system allowing agents to dynamically spawn and coordinate with other agents. Backend: - Add spawn_agent async tool with permission checks - Add OrchestrationConfig model (can_spawn_agents, allowed_agents, max_spawn_depth) - Add GET/PUT /api/agent/orchestration-config endpoint - Use ContextVar pattern for spawn depth tracking - Prevent self-spawning and enforce depth limits Frontend: - Add OrchestrationCard component in Agent Config page - Add provider/model cascade selection in Agent Management modal - Filter providers to show only eligible ones (with API key, models) - Add i18n translations (EN/ZH) Tests: - Add 24 unit tests for spawn_agent tool - Add 11 tests for workflow engine

emoubarak had a problem deploying to maintainer-approved March 22, 2026 01:58 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:02 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:05 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:14 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:25 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:26 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:39 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:40 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:49 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 02:51 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 03:00 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 03:13 — with GitHub Actions Failure

emoubarak had a problem deploying to maintainer-approved March 22, 2026 03:23 — with GitHub Actions Failure

cuiyuebing added this to QwenPaw Mar 25, 2026

github-project-automation Bot moved this to Todo in QwenPaw Mar 25, 2026

emoubarak closed this Mar 31, 2026

github-project-automation Bot moved this from Todo to Done in QwenPaw Mar 31, 2026

Navigation Menu

Repository navigation

feat: Multi-Agent Orchestration System - #2036

feat: Multi-Agent Orchestration System#2036
emoubarak wants to merge 1 commit into
agentscope-ai:mainfrom
emoubarak:feat/workflow-orchestration-engine

emoubarak commented Mar 22, 2026

Uh oh!

This branch had an error being deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

emoubarak commented Mar 22, 2026

Summary

Features

Backend

Frontend

Architecture

Files Changed

Test Coverage

Usage Example

API

Note for Reviewers

Uh oh!

This branch had an error being deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants