Byte IT

ARTICLE SERIES · 19 DEEP DIVES

Superpowers: AI Agent Skills, Deeply Explored

A 19-article series documenting each skill with real examples, patterns, and cross-cutting themes.

Start reading ↓

DEEP-DRILL DEEP-DIVE · 14 ARTICLES

Each skill, explained in depth

Every Superpowers skill as a standalone deep-dive: what it does, when it fires, the real examples from our sessions.

Deep Dive

Brainstorming: Design First

How the brainstorming skill prevents wrong turns by forcing design-first workflows. Includes real examples from Docker infrastructure audits.

Read article
Deep Dive

Writing Plans: Structured Implementation

Turning approved designs into actionable, verifiable task lists with checkpoints and global constraints.

Read article
Deep Dive

Executing Plans: Task-by-Task Delivery

How the executing-plans skill delivers work step by step with checkpoint rules and error handling.

Read article
Deep Dive

Parallel Agents: Concurrent Execution

When parallel agents earn their overhead and when they don't. Real examples from our Docker audits.

Read article
Deep Dive

Systematic Debugging: Root Cause Discipline

The four-phase root cause debugging process: understand, hypothesize, test, verify. No jumping to fixes.

Read article
Deep Dive

Test-Driven Development: RED-GREEN-REFACTOR

Enforcing the Iron Law: write the test before the code. Real examples from Docker and plugin development.

Read article
Deep Dive

Verification Before Completion: Evidence First

Never claim something is fixed without evidence. The verification checklist that catches missed edge cases.

Read article
Deep Dive

Using Superpowers: The Bootstrap Skill

How the using-superpowers skill bootstraps the entire workflow: load all skills, choose the right one, execute.

Read article
Deep Dive

Subagent-Driven Development: Review-Powered Execution

Parallel write agents + review agents working in sync: each PR gets its own dedicated review cycle.

Read article
Deep Dive

Requesting Code Review: Structured Review Requests

How to request code review with context, concerns, and specific questions for each review agent.

Read article
Deep Dive

Receiving Code Review: Technical Evaluation

Treating review feedback as technical evaluation, not performance. Respond with evidence, not agreement.

Read article
Deep Dive

Finishing a Development Branch: The Last 10%

Tests, cleanup, docs, and the discipline of finishing strong. The last 10% that separate shipped from abandoned.

Read article
Deep Dive

Using Git Worktrees: Workspace Isolation

Parallel branches without checkout overhead: each feature gets its own worktree with clean HEAD state.

Read article
Deep Dive

Writing Skills: Meta-Skill Authoring

How each skill is written as a SKILL.md file: frontmatter, structure, and the conventions that make them composable.

Read article

THEME SERIES · 5 ARTICLES

Cross-cutting themes

Themes that connect multiple skills: planning chains, parallelism patterns, verification disciplines, debugging methods, and infrastructure ops.

Theme

Planning to Execution: The Three-Step Chain

How brainstorming → writing-plans → executing-plans chain together. What happens when you skip each step.

Read article
Theme

Parallelism: When Parallel Agents Earn Their Overhead

Not everything should be parallel. The rules for when parallel agents justify their coordination cost.

Read article
Theme

Verification and Quality: Evidence Before Assertions

The patterns that make verification reliable: file hashes, curl checks, grep results, and when each matters.

Read article
Theme

Systematic Debugging: Root-Cause Discipline

Applying systematic debugging across all skills: the four phases and how each skill contributes to each phase.

Read article
Theme

Infrastructure and Ops: Docker and Server Management

How these skills transform Docker/infrastructure work: from ad-hoc commands to structured, verifiable, parallel audits.

Read article

← Back to Byte IT