ORCH — CLI orchestrator for multi-agent teams (complements LlamaIndex multi-agent workflows)

[oxgeneral]

Hey LlamaIndex community! 👋

Built something that complements LlamaIndex's multi-agent capabilities — specifically for teams using CLI-based AI agents alongside their Python workflows.

ORCH is an open-source TypeScript CLI runtime for coordinating teams of AI coding agents:

npm install -g @oxgeneral/orch

# Register CLI agents as team members
orch agent add "architect" --adapter claude --role "Senior developer — implements features"
orch agent add "researcher" --adapter shell --role "Runs analysis scripts and data pipelines"
orch agent add "coder" --adapter codex --role "Writes unit tests"

# Create tasks with dependencies
orch task add "Analyze data structure" -p 1 --assignee researcher
orch task add "Implement feature" -p 2 --assignee architect --depends-on <prev-id>
orch task add "Write tests" -p 3 --assignee coder --depends-on <prev-id>

# Run everything in parallel
orch run --all && orch tui

For LlamaIndex users:

• The shell adapter can run your LlamaIndex Python scripts as ORCH tasks
• ORCH's goal system can coordinate LlamaIndex RAG agents with coding agents
• State machine ensures correct execution order: build index → query → validate → done

Key features:

• Typed state machine: todo → in_progress → review → done with auto-retry on failures
• Scope locking — prevents concurrent agents from editing the same files
• Inter-agent messaging — pass query results and context between agents without copy-paste
• Real-time TUI dashboard — visibility across all running agents
• 1,657 tests · TypeScript strict · MIT · zero-database

GitHub: https://github.com/oxgeneral/ORCH
npm: npm install -g @oxgeneral/orch

Would love feedback from LlamaIndex users — especially around bridging Python agentic workflows with CLI-based coding agents!