orchestrate

Decompose large goals into parallel Cursor cloud-agent worker trees with structured handoffs.

Updated May 18, 2025
One-click install
npx skills add https://github.com/nthpaul/dotfiles --skill orchestrate-nthpaul
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: orchestrate
Source: https://github.com/nthpaul/dotfiles/tree/main/cursor/.cursor/plugins/cache/cursor-public/orchestrate/d1cdb88a9eb33cf392395c87e3fd76419fc1010e/skills/orchestrate
Command: npx skills add https://github.com/nthpaul/dotfiles --skill orchestrate-nthpaul

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @cursor/sdk, @slack/web-api, commander, zod, zod-to-json-schema, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill eliminates the difficulty of decomposing a large work item into coordinated parallel Cursor cloud-agent tasks and safely collecting structured handoffs back into a single plan-driven workflow.

Core Features & Use Cases

  • Parallel cloud-agent decomposition: Uses a dispatcher/planner model to fan out a user goal across a tree of planners, workers, verifiers, and merges.
  • Structured handoffs as the only data bus: Workers and verifiers end with a strict handoff structure so the planner can reliably decide what comes next.
  • Verification-driven quality gates: Supports verifier verdicts (live UI, unit tests, type-check only, or blocked/failed) to avoid shipping unproven fixes.
  • Operational guardrails: Provides Andon to pause spawning, failure-mode sidecars for crashed runs, and loop hygiene/checkpoint restarts for robustness.
  • Optional Slack run visibility: Mirrors task status in a single run thread when Slack credentials are present, without changing correctness when Slack is unavailable.

Quick Start

Use the orchestrate skill by typing /orchestrate <goal> to decompose and execute the goal across parallel cloud agents, with results collected via structured handoffs.

Frequently Asked Questions about orchestrate

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

FAQPage Schema
How do I decompose a large software engineering goal into parallel Cursor agents?

To decompose a large goal into parallel Cursor agents, use a planner-worker-verifier handoff protocol that fans out tasks across isolated cloud-agents, collecting structured handoffs to drive iterative planning without cross-talk between workers.

What is structured handoff protocol in parallel agent orchestration?

Structured handoffs in parallel agent orchestration are the strict data format workers and verifiers use to pass results back to the planner, ensuring reliable decisions on what comes next and preventing unproven fixes from shipping.

How do I run verification checks on parallel Cursor cloud-agent tasks?

You can run verification checks on parallel Cursor agents by using verifier verdicts like live UI, unit tests, type-check only, or blocked/failed, which act as quality gates to ensure only proven code passes the planner.

Do I need the Cursor SDK to fan out parallel agents?

Yes, you need the Cursor SDK integration for spawning and authentication to fan out parallel agents, as it provides the underlying infrastructure to distribute work across isolated cloud-agent workers and collect their structured handoffs.

Can I mirror parallel agent task status in Slack?

Yes, you can mirror task status in a single Slack run thread when Slack credentials are present, providing optional run visibility without changing the correctness of the agent orchestration workflow if Slack is unavailable.

How do I handle crashed parallel agent runs during task orchestration?

To handle crashed parallel agent runs, use operational guardrails like Andon to pause spawning, failure-mode sidecars for crashed runs, and loop hygiene with checkpoint restarts to maintain robustness during complex orchestration.