galangal-orchestrate 0.57.2

AI-driven development workflow orchestrator

Galangal Orchestrate

Turn AI coding assistants into structured development workflows.

Galangal wraps Claude Code CLI to execute a deterministic, multi-stage development pipeline with approval gates, validation, and automatic rollback.

Why Use This?

Instead of open-ended AI coding sessions, you get a structured workflow:

1. PM - AI writes requirements, you approve before code is written
2. Design - AI proposes architecture, you approve the approach
3. Dev - AI implements according to approved specs
4. Test/QA/Review - Automated validation with rollback on failure
5. Docs - AI updates documentation

If anything fails, the workflow rolls back with context about what went wrong.

Quick Start

# Install (lean core)
pip install galangal-orchestrate

# Optional: semantic mistake tracking (adds sentence-transformers/torch, ~hundreds of MB)
pip install "galangal-orchestrate[full]"

# Initialize in your project
cd your-project
galangal init

# Start a task
galangal start "Add user authentication with JWT"

# Run non-interactively (CI/scripts, no TUI)
galangal start "Fix the login bug" --type bugfix --headless

# Check status / resume
galangal status
galangal resume

Requirements

• Python 3.10+
• Claude Code CLI (claude command)
• Git

Commands

Command	Description
galangal init	Initialize in current project
galangal start "desc"	Start new task
galangal status	Show task status
galangal resume	Continue active task
galangal list	List all tasks
galangal index stats	Show task index DB stats
galangal index rebuild	Rebuild task index from task folders
galangal index migrate-artifacts	Import legacy artifact files into DB and delete file copies
galangal index compact-done	Keep only PLAN.md and SUMMARY.md in tasks/done/*
galangal complete	Finalize & create PR

Interactive controls during execution:

• ^Q Quit/pause
• ^I Interrupt with feedback
• ^N Skip stage
• ^B Go back
• ^E Pause for manual edit

Galangal Hub

Monitor and control workflows remotely across multiple machines.

# Deploy hub server (Docker)
docker run -d -p 8080:8080 \
  -e HUB_USERNAME=admin \
  -e HUB_PASSWORD=your-password \
  -e HUB_API_KEY=your-api-key \
  -v galangal-hub-data:/data \
  ghcr.io/galangal-media/galangal-hub:latest

# Enable in your project (.galangal/config.yaml)
hub:
  enabled: true
  url: ws://your-server:8080/ws/agent
  api_key: your-api-key  # Must match HUB_API_KEY on server

See Hub Documentation for full setup instructions.

Documentation

Topic	Link
Getting Started	docs/getting-started.md
Configuration	docs/guide/configuration.md
Workflow Stages	docs/guide/workflow-pipeline.md
Hub (Remote Control)	docs/hub/README.md
Troubleshooting	docs/troubleshooting.md
Architecture	docs/local-development/architecture.md

Task Types

Type	Stages	Use Case
Feature	All stages	New functionality
Bug Fix	PM → PREFLIGHT → DEV → TEST → TEST_GATE → QA → REVIEW → SUMMARY	Fixing bugs
Refactor	PM → DESIGN → PREFLIGHT → DEV → TEST → TEST_GATE → REVIEW → SUMMARY	Code restructuring
Chore	PM → PREFLIGHT → DEV → TEST → TEST_GATE → REVIEW → SUMMARY	Config, dependencies
Docs	PM → DOCS → SUMMARY	Documentation only
Hotfix	PM → DEV → TEST → TEST_GATE → SUMMARY	Critical fixes

License

MIT License - see LICENSE file.