orchestration_skill

This skill orchestrates multiple agents to plan, delegate, and track complex tasks, delivering parallel subtasks and real-time session visibility.
  • TypeScript

868

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill numman-ali/n-skills --skill orchestration

  • SKILL.md35.6 KB

Overview

This skill orchestrates multi-agent work for complex engineering tasks by decomposing goals, tracking dependencies, and providing real-time session visibility. It combines persistent task graphs (cc-mirror tasks) with live tactical updates (TodoWrite) so large efforts run in parallel without losing order. Use it to turn high-level requests into coordinated, trackable workstreams.

How this skill works

The orchestrator reads domain guidance, breaks the request into independent and dependent work items, and creates persistent tasks with cc-mirror. It mirrors task state into TodoWrite using a simple icon protocol so the user sees live progress. Worker agents execute delegated tasks, report results, and the orchestrator resolves tasks and synthesizes outputs.

When to use it

  • Building or shipping new features that decompose into parallel workstreams
  • Refactors, large code changes, or migrations with clear dependency chains
  • Comprehensive PR reviews, security audits, or multi-part testing efforts
  • Documentation, release notes, or cross-team deliverables requiring coordination
  • Any request where persistent dependency tracking and live session visibility matter

Best practices

  • Always load the relevant domain guide before decomposing work
  • Create cc-mirror tasks for each discrete work item and encode dependencies immediately
  • Use TodoWrite to expose only the session-relevant view; hide orchestration machinery from the user
  • Spawn worker agents for heavy reading or multi-file analysis; limit direct reads to 1–2 files
  • Keep worker prompts explicit: execute the task, use tools, report results, and stop

Example use cases

  • Split a feature request into backend, frontend, tests, and docs tasks, run them in parallel, and link blockers
  • Run a security review by spawning reviewers for different modules and aggregating their findings
  • Coordinate a migration: create migration tasks, mark DB tasks as blocking app updates, and show live progress
  • Generate and run test suites in parallel while tracking which tests block release readiness
  • Produce a release: spawn docs, changelog, CI config, and deployment workers and synthesize final release notes

FAQ

The orchestrator reads a few domain files and the skill references; heavyweight reading or code exploration is delegated to worker agents.

How are dependencies displayed to the user?

TodoWrite entries use icons (○ open, ● blocked, ✓ completed, ⚠ has blockers) and list blocker IDs so users see actionable state without orchestration details.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational