Verified:GitHub

projectStart

Transforms PRD tasks into a dependency DAG with partitioned segments for parallel execution.

Updated May 21, 2026

Search Tags:#dependency management #bounded context #task planning #ai delegation #dag #project orchestration #file ownership

Ready to run

npx skills add https://github.com/kinetas/harness_engineering --skill projectstart

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: projectStart
Source: https://github.com/kinetas/harness_engineering/tree/main/projectStart
Command: npx skills add https://github.com/kinetas/harness_engineering --skill projectstart

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Complex project execution often stalls because tasks lack clear dependencies, ownership boundaries, and an orchestrated plan for parallel work across teams and AI agents.

Core Features & Use Cases

Builds a dependency-aware DAG: Converts PRD tasks into a directed acyclic graph with explicit dependency edges.
Partitions work into dependency islands: Groups tasks into segments using topological layering and domain clustering while enforcing file ownership to prevent conflicts.
Delegates segments to Manager AIs with contracts: Spawns bounded-context coordinators in an execution order driven by imports/exports dependency contracts and coordinated Sub AI capacity limits.
Produces a start-up report: Triggers a Collector AI update to publish the segment/task plan in report/report.md immediately after kickoff.

Quick Start

Issue the command /projectStart after ensuring doc/PRD.md contains real project requirements and doc/company_state.json reflects the current system state.

Frequently Asked Questions about projectStart

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

FAQPage Schema

How do I plan parallel task execution for a full project kickoff?▼

To plan parallel task execution for a full project kickoff, transform PRD tasks into a dependency DAG and partition them into dependency-island segments using topological layering. This coordinates backend, frontend, and infra tasks while enforcing strict file ownership isolation.

What is dependency-island segmentation in project orchestration?▼

Dependency-island segmentation is a project orchestration technique that groups tasks into bounded segments using topological layering and domain clustering. It enforces file ownership rules to prevent conflicts and meets imports/exports dependency contract constraints with bounded manager spawn limits.

How do I coordinate parallel AI delegation across multiple bounded contexts?▼

To coordinate parallel AI delegation, spawn bounded-context coordinators as Manager AIs in an execution order driven by imports/exports dependency contracts. This ensures strict file ownership isolation and respects bounded sub-AI capacity limits across backend, frontend, infra, testing, and docs tasks.

Do I need a PRD to generate a dependency DAG for task planning?▼

Yes, you need a PRD to generate a dependency DAG for task planning. You must issue the kickoff command after ensuring doc/PRD.md contains real project requirements and doc/company_state.json reflects the current system state to build an accurate execution plan.

How does topological layering with BFS handle file ownership rules in a DAG?▼

Topological layering with BFS handles file ownership rules in a DAG by partitioning tasks into dependency-island segments. This deterministic segmentation enforces strict file ownership isolation, preventing conflicts when delegating bounded-context segments to Manager AIs.

How do I get a startup report after launching a project kickoff?▼

To get a startup report after launching a project kickoff, trigger a Collector AI update to publish the segment and task plan in report/report.md. This happens immediately after the dependency DAG and parallel manager delegation execution order are established.