Verified:GitHub

subagent-driven-development

Orchestrate subagent worker waves with write-set ownership and verification gates.

Updated Apr 27, 2026

Search Tags:#code execution #subagent #plan orchestration #goal completion #workflow receipts #write-set safety #parallel task waves

Ready to run

npx skills add https://github.com/vTRKA/supervibe --skill subagent-driven-development-vtrka

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: subagent-driven-development
Source: https://github.com/vTRKA/supervibe/tree/main/skills/subagent-driven-development
Command: npx skills add https://github.com/vTRKA/supervibe --skill subagent-driven-development-vtrka

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents confusing, single-threaded AI coding sessions by turning an approved execution plan into a wave of independent, verifiable worker tasks, each with explicit scope, evidence, and stop/resume controls.

Core Features & Use Cases

Parallel subagent execution: Fan out 5+ independent tasks as bounded worker packets so progress doesn’t stall on one checkbox-style win.
Write-set ownership & conflict safety: Enforce disjoint write sets and block dispatch when overlap or missing ownership evidence appears.
Verification and durable receipts: Require per-task verification commands and runtime host invocation proof with receipts before counting work as complete.
Final-sweep review policy: Run reviewer/verification coverage once at the final graph/epic/release sweep, not mid-task by default.

Quick Start

Run a /supervibe subagent-driven execution only after you have an approved loop-ready plan or accepted work graph and the latest handoff question is answered in the user gate.

Frequently Asked Questions about subagent-driven-development

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

FAQPage Schema

How do I execute parallel coding tasks safely without file edit conflicts?▼

Parallel coding tasks require write-set ownership enforcement to prevent file edit conflicts. This approach atomizes an approved plan into bounded worker waves, assigning disjoint write sets to independent subagents so concurrent edits remain safe and verifiable.

What is subagent-driven plan orchestration for software development?▼

Subagent-driven plan orchestration is the process of fanning out independent implementation tasks as bounded worker packets. It assigns Goal IDs and verification gates to each packet, ensuring parallel execution remains safe under concurrent edits.

How do I start orchestrating subagents for an approved implementation plan?▼

Orchestrating subagents requires an approved loop-ready plan or accepted work graph first. You then dispatch bounded worker waves with explicit scope, physical subagent profile binding, and stop/resume controls to execute independent tasks in parallel.

Can I use parallel subagents for multi-task epics where independent tasks run concurrently?▼

Yes, parallel subagents are designed for multi-task epic execution scenarios where independent tasks run concurrently. The system fans out 5+ independent tasks as bounded worker packets, enforcing disjoint write sets to block dispatch when overlap appears.

Why does concurrent AI coding stall during single-threaded sessions?▼

Concurrent AI coding stalls in single-threaded sessions because progress halts on individual checkbox-style wins. Fanning out independent tasks as bounded worker waves with verification gates prevents this bottleneck and enables continuous parallel progress.

When should I run final-sweep verification coverage for parallel subagent tasks?▼

Final-sweep verification coverage should run once at the final graph, epic, or release sweep, not mid-task by default. This policy requires per-task verification commands and runtime host invocation proof with durable receipts before counting work complete.