phantom

Dispatch fresh subagents per task and enforce two-stage spec and code reviews.

67|10|Updated Apr 12, 2026
One-click install
npx skills add https://github.com/GadaaLabs/claude-code-on-steroids --skill phantom-gadaalabs
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: phantom
Source: https://github.com/GadaaLabs/claude-code-on-steroids/tree/main/skills/phantom
Command: npx skills add https://github.com/GadaaLabs/claude-code-on-steroids --skill phantom-gadaalabs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Executes each implementation task through a fresh subagent with zero session context, then performs a two-stage review (spec compliance → code quality) before accepting results, enabling isolation and faster iteration without context pollution.

Core Features & Use Cases

  • Fresh subagent per task to avoid cross-task contamination and preserve session state.
  • Two-stage review after each task: first verify spec compliance, then assess code quality.
  • Per-task isolation enables parallel execution (EXODUS) and robust guardrails across complex plans.
  • Clear handoff workflow for implementer, spec reviewer, and code quality reviewer, culminating in a final code review.

Quick Start

Dispatch a fresh subagent for each task from your plan and apply a two-stage review (spec compliance → code quality) before accepting results.

Frequently Asked Questions about phantom

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

FAQPage Schema
How do I isolate tasks with fresh subagents to avoid context pollution?

Spec compliance is verified by running a dedicated review stage that checks the subagent's output against the original specifications. This ensures spec fidelity before the results are accepted into your codebase.

How do I enforce a two-stage review for code quality and spec compliance?

A two-stage review enforces code quality by first verifying spec compliance, then assessing code quality through a separate reviewer prompt. This structured handoff workflow ensures both accuracy and robustness before accepting results.

Can I use per-task isolation to enable parallel execution for complex plans?

You need a defined implementation plan and templated workflow with implementer and reviewer prompts. The skill provides these templates to guide end-to-end execution, culminating in a final code review.

What is the best way to prevent cross-task contamination when automating code generation?

The best way to prevent cross-task contamination is dispatching a fresh subagent for each implementation task. This isolation ensures zero session context bleeds between tasks, maintaining clean guardrails and preserving session state.

Why does task isolation matter for maintaining code quality in automated workflows?

Task isolation matters for code quality because it enforces a clean execution environment with zero session context for each subagent. This prevents cross-task contamination and supports a structured two-stage review process.