dispatching-parallel-agents

Dispatches native subagents across independent work lanes within Spec Kit Plus workflows.

Updated Jul 6, 2026
One-click install
npx skills add https://github.com/chenziyang110/launchdeck --skill dispatching-parallel-agents-chenziyang110
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dispatching-parallel-agents
Source: https://github.com/chenziyang110/launchdeck/tree/main/.specify/templates/passive-skills/dispatching-parallel-agents
Command: npx skills add https://github.com/chenziyang110/launchdeck --skill dispatching-parallel-agents-chenziyang110

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When facing multiple independent test failures, bugs, or implementation tasks, running them serially wastes time, while ad-hoc parallelization risks write conflicts and unverifiable results. This Skill provides a disciplined routing and dispatch process for parallel subagent execution. ## Core Features & Use Cases - Lane-Based Decomposition: Splits work into named lanes with explicit owners, write-sets, forbidden paths, and verification targets before dispatch. - Native Subagent Dispatch: Routes through owning sp-* workflows (sp-quick, sp-debug, sp-map-scan, sp-map-build, sp-implement) and dispatches via native subagent surfaces like Codex spawn_agent/wait_agent or Claude Task. - Conflict and Fallback Rules: Prevents concurrent writers on shared state and defines fallbacks such as sp-teams or sequential execution when lanes are coupled. - Use Case: A developer faces five failing tests spread across unrelated modules; the Skill splits them into independent lanes, dispatches parallel subagents, and joins results through structured handoffs listing changed files and verification output. ## Quick Start Ask the agent to dispatch parallel subagents for the independent failing tests or tasks, routing through the appropriate sp-* workflow with defined write-sets and verification targets.

Frequently Asked Questions about dispatching-parallel-agents

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

FAQPage Schema
How do I run multiple independent tasks in parallel with AI subagents?▼

Split the work into named lanes with explicit write-sets, forbidden paths, and verification targets, then dispatch native subagents such as Codex spawn_agent/wait_agent or Claude Task. Join results only after each worker reports changed files and verification output.

When should I use parallel subagents instead of sequential execution?▼

Use parallel subagents when two or more lanes are independent, have non-overlapping write-sets, and parallelism materially improves throughput or evidence quality. If lanes share write state or one fix may resolve others, keep the work sequential in one workflow run.

Can parallel agents edit the same files concurrently?▼

No. The Skill forbids dispatching two writers to the same file or shared state unless one lane is explicitly read-only or the owning workflow defines a safe join point. Conflicting lanes must be sequenced or merged under one owner.

What happens if native subagents are not available in my runtime?▼

Fall back to sp-teams when durable team state and explicit join tracking are needed, or use a separate terminal only when no native subagents or managed-team path exist. Opening another terminal is never the first option.

Why should a worker prompt include write-sets and verification targets?▼

Without a defined write-set, forbidden paths, and verification target, workers can conflict with other lanes or return unverifiable results. A handoff saying only "done" without file paths and command output is rejected.