project-orchestrator

Route user requests to skills and decompose work into coordinated subagents.

3|Updated Apr 5, 2026
One-click install
npx skills add https://github.com/dvy1987/agent-loom --skill project-orchestrator-dvy1987
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: project-orchestrator
Source: https://github.com/dvy1987/agent-loom/tree/main/.agents/skills/project-orchestrator
Command: npx skills add https://github.com/dvy1987/agent-loom --skill project-orchestrator-dvy1987

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Projects often span many specialised tasks and skills, and users need a single coordinator to decide what runs next, which specialist to call, and whether work can safely run in parallel across available platforms. The Project Orchestrator reduces guesswork, prevents unsafe parallelism, and ensures phase-appropriate routing so teams and agents execute the right steps in the right order.

Core Features & Use Cases

  • Phase detection: Silent scans repository artifacts to determine the current project phase and recommend the next skill to run.
  • Skill routing & delegation: Maps user intent to a single skill or a decomposed process and never executes specialist work directly.
  • Platform-aware orchestration: Adapts execution strategy to platform capabilities (Tier 1 parallel subagents, Tier 2 platform-managed dispatch, Tier 3 sequential execution).
  • Parallel decomposition rules & safety: Enforces file-scope and dependency rules to prevent conflicts when parallelising tasks.
  • Process learning loop: Records execution outcomes, updates process entries, and triggers AGENTS.md refreshes only when architectural or orchestration maps change.
  • Use Case: Turn a high-level "implement feature X" request into a phased plan that runs product-soul → brainstorming → prd-writing → implementation-plan, with scoped parallel subagents where safe.

Quick Start

Ask the orchestrator to "Orchestrate this project and propose a phase-by-phase plan based on repository state."

Frequently Asked Questions about project-orchestrator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate subagents for parallel execution in a multi-phase project?

Project orchestration coordinates subagents by decomposing complex work into phases, enforcing file-scope rules, and routing tasks to prevent unsafe parallelism. It scans repository artifacts to determine the current phase and presents an execution plan for parallel or sequential dispatch.

What is skill routing and how does it manage process decomposition?

Skill routing maps user intent to specialized skills without executing specialist work directly. Process decomposition breaks complex requests into coordinated steps, selecting sequential or parallel execution based on platform capabilities and dependency rules to ensure safe task isolation.

How do I coordinate project phases when my repository needs sequential and parallel task execution?

Platform-aware orchestration adapts execution strategy to platform capabilities, using Tier 1 parallel subagents, Tier 2 platform-managed dispatch, or Tier 3 sequential execution. It performs silent repository scans to detect the project phase and recommend the next skill to run.

Can I use parallel decomposition rules to prevent file-scope conflicts across subagents?

Yes, parallel decomposition enforces file-scope and dependency rules to prevent conflicts when parallelising tasks. The orchestrator presents an execution plan for approval, ensuring subagents operate within safe boundaries and architectural constraints are respected.

Does project orchestration work with platform-managed dispatch for sequential execution?

Platform-aware orchestration adapts to available platform capabilities, falling back to Tier 2 platform-managed dispatch or Tier 3 sequential execution when parallel subagents are not supported. This ensures phase-appropriate routing regardless of platform constraints.

How does the process learning loop update orchestration maps after execution?

The process learning loop records execution outcomes and updates process entries. It triggers AGENTS.md refreshes only when architectural or orchestration maps change, keeping the repository's orchestration metadata synchronized with actual project state.