From emasoft-orchestrator-agent
Use when managing orchestration phase commands. Trigger with start, monitor, loop control, cancellation, or stop hook enforcement requests.
npx claudepluginhub emasoft/emasoft-plugins --plugin emasoft-orchestrator-agentThis skill uses the workspace's default tool permissions.
This skill teaches how to use the orchestration phase commands in the Orchestrator Agent plugin. These commands manage the execution loop that coordinates remote agents to implement approved plans.
README.mdreferences/cancellation-cleanup.mdreferences/op-cancel-orchestrator.mdreferences/op-check-orchestrator-status.mdreferences/op-monitor-orchestration-status.mdreferences/op-start-orchestration.mdreferences/op-start-orchestrator-loop.mdreferences/op-stop-hook-enforcement.mdreferences/orchestration-loop-mechanics.mdreferences/start-orchestration-procedure.mdreferences/state-file-format.mdreferences/status-monitoring.mdreferences/troubleshooting.mdRuns orchestration commands to coordinate AI agents executing approved plans. Triggers: /start-orchestration, /orchestration-status, /cancel-orchestrator. Use post-plan approval.
Automates multi-agent execution with task decomposition, parallel dispatch, memory coordination, and verification loops for complex features.
Orchestrates multi-agent workflows with parallel pipelines, sync barriers, state tracking, checkpointing, and progress metrics. Use for coordinating 3+ agents across sessions.
Share bugs, ideas, or general feedback.
This skill teaches how to use the orchestration phase commands in the Orchestrator Agent plugin. These commands manage the execution loop that coordinates remote agents to implement approved plans.
| Output Type | Format | Example |
|---|---|---|
| Status report | Markdown table | Module completion percentages, agent assignments |
| Loop state | Text summary | Active/inactive, iteration count, pending tasks |
| Cancellation | Confirmation | "Orchestrator loop cancelled at iteration X" |
| Error messages | Text | Hook blocking reasons, state file issues |
Before using orchestration commands, ensure:
/approve-plan)design/state/exec-phase.md exists| Command | Purpose | When to Use |
|---|---|---|
/start-orchestration | Activate orchestration phase | After plan approval, to begin implementation |
/orchestration-status | View module/agent progress | To monitor implementation progress |
/orchestrator-status | Check loop state and tasks | To see pending tasks across all sources |
/orchestrator-loop | Start continuous task loop | To activate task-driven development loop |
/cancel-orchestrator | Cancel active loop | To manually stop the orchestrator loop |
# Step 1: Start orchestration with GitHub Project sync
/start-orchestration --project-id PVT_kwDOB1234567
# Step 2: Register your agents
/register-agent ai implementer-1 --session helper-agent-generic
# Step 3: Assign first module
/assign-module auth-core implementer-1
# Step 4: Monitor progress
/orchestration-status
# Start continuous task-driven development
/orchestrator-loop "Complete all pending authentication tasks" --max-iterations 50
# Check loop status
/orchestrator-status --verbose
# Cancel if needed
/cancel-orchestrator
If you need to start the orchestration phase after plan approval:
/start-orchestration (optionally with --project-id for GitHub sync)/register-agent/assign-module/orchestration-status every 10-15 minutesSee references/start-orchestration-procedure.md:
If you need to check implementation progress:
Use /orchestration-status to see:
See references/status-monitoring.md:
If you need to start or manage the orchestrator loop:
The orchestrator loop monitors multiple task sources and prevents exit until ALL are complete:
See references/orchestration-loop-mechanics.md:
If you need to cancel the orchestrator loop:
Use /cancel-orchestrator to:
See references/cancellation-cleanup.md:
If you need to understand or debug state files:
The orchestrator uses markdown files with YAML frontmatter to track state:
design/state/loop.md - Loop statedesign/state/exec-phase.md - Execution phase stateSee references/state-file-format.md:
See references/troubleshooting.md:
The following scripts implement the orchestration commands. Located in the plugin's scripts/ directory:
| Script | Purpose | Used By |
|---|---|---|
eoa_start_orchestration.py | Activates orchestration phase | /start-orchestration |
eoa_orchestration_status.py | Displays phase status | /orchestration-status |
eoa_check_orchestrator_status.py | Shows loop state | /orchestrator-status |
eoa_setup_orchestrator_loop.py | Creates loop state file | /orchestrator-loop |
eoa_orchestrator_stop_check.py | Stop hook enforcement | Hook event |
Purpose: Enter Orchestration Phase to coordinate remote agents implementing the approved plan.
Syntax:
/start-orchestration [--project-id PVT_kwDOBxxxxxx]
Options:
--project-id: GitHub Project ID for Kanban synchronizationWhat it does:
Example:
/start-orchestration --project-id PVT_kwDOB1234567
Purpose: View current Orchestration Phase progress including modules, agents, and verification status.
Syntax:
/orchestration-status [--verbose] [--agents-only] [--modules-only]
Options:
--verbose: Show detailed polling history and acceptance criteria--agents-only: Show only agent information--modules-only: Show only module statusOutput sections:
Purpose: Check orchestrator loop status and pending tasks across all sources.
Syntax:
/orchestrator-status [--verbose]
Options:
--verbose: Show detailed debug information and recent log entriesOutput sections:
Purpose: Start orchestrator loop for continuous task-driven development.
Syntax:
/orchestrator-loop [PROMPT] [options]
Options:
--max-iterations N: Maximum iterations before escalation (default: 100)--completion-promise TEXT: Promise phrase to trigger completion--task-file PATH: Markdown task file to monitor--check-tasks BOOL: Check Claude Tasks (default: true)--check-github BOOL: Check GitHub Projects (default: true)--github-project ID: Specific GitHub Project IDExample:
/orchestrator-loop "Complete all pending authentication tasks" --max-iterations 50
Purpose: Cancel active orchestrator loop.
What it does:
Note: This command is hidden from slash command suggestions but remains functional.
The orchestrator stop hook (eoa_orchestrator_stop_check.py ) enforces completion requirements:
Blocking conditions:
Completion signals:
ALL_TASKS_COMPLETE when all tasks genuinely done<promise>YOUR_PHRASE</promise> matching configured promiseRecovery behavior:
# Step 1: Start orchestration with GitHub Project sync
/start-orchestration --project-id PVT_kwDOB1234567
# Step 2: Register AI agent
/register-agent ai implementer-1 --session helper-agent-generic
# Step 3: Assign first module
/assign-module auth-core implementer-1
# Step 4: Monitor progress
/orchestration-status --verbose
# Start loop
/orchestrator-loop "Complete all authentication tasks" --max-iterations 50
# Check status
/orchestrator-status --verbose
# If needed, cancel
/cancel-orchestrator
# Check all agents every 10-15 minutes
/check-agents
# Or check specific agent
/check-agents --agent implementer-1
# View orchestration status
/orchestration-status --modules-only
Copy this checklist and track your progress:
/approve-plan executed)design/state/exec-phase.md exists/start-orchestration with optional --project-id/register-agent ai <agent_id> --session <session>/assign-module <module_id> <agent_id>/check-agents every 10-15 minutesCopy this checklist and track your progress:
/orchestration-status to see module completion--verbose flag for detailed diagnostics/orchestrator-status for pending tasks across sourcesCopy this checklist and track your progress:
/cancel-orchestrator.claude/design/state/loop.md