Agent skill
orchestrate
Task Decomposition Engine - Breaks down complex tasks into phases, creates Native Tasks with dependencies, and generates worker prompts. Pipeline Position: After /planning, Before /assign Handoff: /assign --workload {slug}
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/orchestrate
SKILL.md
/orchestrate - Task Decomposition Engine
Version: 4.0.0 | Model: opus Pipeline: /planning -> [/orchestrate] -> /assign EFL: P1-P6 Complete
1. Purpose
Task Decomposition Engine that:
- Breaks down complex tasks into phases
- Creates Native Tasks via
TaskCreate - Sets up dependencies via
TaskUpdate(addBlockedBy) - Generates worker prompt files (
.agent/prompts/{slug}/pending/) - Initializes workload-specific context and progress tracking
Does NOT:
- Assign tasks to terminals (use
/assign) - Execute tasks (handled by
/worker) - Set
ownerfield (remains null until/assign)
2. Task API Integration
Task Creation Pattern
// Create tasks in dependency order
for (phase of analysis.phases) {
task = TaskCreate({
subject: phase.name,
description: phase.description,
activeForm: `Working on ${phase.name}`,
metadata: {
phaseId: phase.id,
priority: phase.priority || "P1",
promptFile: `${workloadPromptDir}/pending/${filename}`
}
})
taskMap[phase.id] = task.id
}
// Set up dependencies
for (phase of analysis.phases) {
if (phase.dependencies.length > 0) {
TaskUpdate({
taskId: taskMap[phase.id],
addBlockedBy: phase.dependencies.map(d => taskMap[d])
})
}
}
Output Files
| File | Purpose |
|---|---|
_context.yaml |
Project context and phase definitions |
_progress.yaml |
Progress tracking per terminal |
pending/*.yaml |
Worker task prompts |
3. Invocation
# With plan slug (from /planning)
/orchestrate --plan-slug user-auth-20260128
# Standalone (auto-generates workload)
/orchestrate "Implement user authentication system"
4. Execution Protocol
Phase 1: Requirements Analysis
// Parse input and generate workload ID
const workloadId = generateWorkloadId(projectSlug)
const workloadSlug = generateSlugFromWorkload(workloadId)
// Analyze task for decomposition
const analysis = analyzeTask(input)
// Returns: { project, objectives, phases[], estimatedWorkers }
Phase 2: Gate 4 Validation (Shift-Left)
// Validate phase dependencies BEFORE creating tasks
const gate4Result = await validatePhaseDependencies(phases)
if (gate4Result.hasErrors) {
return { status: "gate4_failed", errors: gate4Result.errors }
}
// Log warnings but continue
if (gate4Result.hasWarnings) {
logValidationWarnings("ORCHESTRATE", gate4Result.warnings)
}
Phase 3: Initialize Workload Directory
// Create workload structure
Bash(`source .claude/skills/shared/workload-tracker.sh && init_workload_directories "${workloadId}"`)
Bash(`source .claude/skills/shared/workload-files.sh && set_active_workload "${workloadId}"`)
Phase 4: Native Task Creation
// Create tasks with TaskCreate
for (phase of phases) {
const task = TaskCreate({
subject: phase.name,
description: phase.description,
activeForm: `Working on ${phase.name}`
})
taskMap[phase.id] = task.id
}
Phase 5: Dependency Setup
// Set blockers with TaskUpdate
for (phase of phases) {
if (phase.dependencies.length > 0) {
TaskUpdate({
taskId: taskMap[phase.id],
addBlockedBy: phase.dependencies.map(d => taskMap[d])
})
}
}
Phase 6: Generate Context Files
_context.yaml: Project metadata, phases, dependency graph_progress.yaml: Terminal status, phase progress
Phase 7: Generate Worker Prompts
- Create
pending/worker-phase-{n}-{slug}-task.yamlfor each phase - Include scope, target files, completion criteria, validation
Phase 8: Summary Output
=== Orchestration Complete ===
Project: {project}
Tasks: {count}
Next: /assign auto
5. Error Handling
| Error | Recovery |
|---|---|
| Invalid input | Prompt for clarification |
| Too many phases (>10) | Suggest sub-orchestration |
| Circular dependency | Reject, show cycle |
| Gate 4 failed | Show errors, abort |
6. Handoff Contract
# On completion, output:
handoff:
skill: "orchestrate"
workload_slug: "{slug}"
status: "completed"
next_action:
skill: "/assign"
arguments: "--workload {slug}"
7. Version History
| Version | Changes |
|---|---|
| 4.0.0 | Deduplicated, V2.1.19 frontmatter, Task API patterns |
| 3.0.0 | Full EFL P1-P6 implementation |
| 2.1.0 | V2.1.19 Spec compatibility |
| 1.0.0 | Initial task orchestration |
End of Skill Documentation
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
majiayu000/claude-skill-registry
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
majiayu000/claude-skill-registry
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?