Open source · Claude Code plugin

20 skills that turn Claude Code into a senior engineering partner.

Born from reverse-engineering how top-tier AI coding agents actually work internally: session management, multi-agent orchestration, adversarial testing, decision archaeology, calibrated estimation. Then packaging those patterns as drop-in skills. No wrappers. No frameworks. No dependencies.

Why Claude Power Skills exists

Claude Code ships with solid defaults. But if you have pushed it on real work (multi-repo monoliths, production incident debugging, sprint planning), you have hit the ceiling. The agent forgets context between sessions. It estimates effort like someone who has never shipped software. It does not know when it is stuck. It cannot coordinate parallel workstreams.

These skills fix that. Each one encodes a specific workflow pattern that experienced engineers use intuitively, structured so the agent can execute it consistently.

Things you can ask

/bughunter — adversarial security analysis of this codebase.

/ultraplan — deep planning with creative exploration before we build.

/effort-estimate — calibrated estimate using my past velocity.

/codebase-insight — persistent architecture analysis.

/thinkback — reconstruct why this decision was made.

How to install Claude Power Skills

Install as a plugin (recommended):

/plugin install irfad7/claude-power-skills

That is it. All 20 skills are available immediately across all your projects.

Or cherry-pick into a single project:

git clone https://github.com/irfad7/claude-power-skills.git ~/claude-power-skills
cp -r ~/claude-power-skills/skills/bughunter your-project/skills/
cp -r ~/claude-power-skills/skills/ultraplan your-project/skills/

Once installed, invoke by name:

/bughunter        → adversarial security analysis
/ultraplan        → deep planning with creative exploration
/effort-estimate  → calibrated estimation with velocity tracking
/codebase-insight → persistent architecture analysis

What's inside the repo

SkillWhat it does
kairos-sessionSession lifecycle — boot sequence, context loading, state diffing, memory consolidation on shutdown.
autodreamMemory synthesis across sessions. Detects contradictions, merges observations, outputs a health score.
session-reviewEnd-of-session analysis. Extracts decisions, computes knowledge deltas, writes handoff notes.
context-compressionThree-layer compression for long conversations, from ~30% lossless up to ~90%.
ultraplanFive-act planning: understand, explore, validate, design, phase. With devil's advocate and YAGNI audits.
effort-estimateCalibrated estimation. Decomposes to atoms, scores complexity, calibrates on your past velocity.
thinkbackDecision archaeology — reconstruct why a past decision was made from git history and code.
bughunterAdversarial bug hunting.
security-reviewSecurity audit pass.
code-reviewStructured code review.
coordinator-modeMulti-agent orchestration across parallel workstreams.
+ 9 morebatch-mode, prompt-speculation, pr-autofix, frustration-aware, anti-distillation, codebase-insight, magic-docs, stuck-recovery, undercover-mode.

Frequently asked questions

What are Claude Code skills?

Skills are Markdown files (SKILL.md) that Claude Code loads to learn a specific workflow. When a task matches a skill's description, Claude reads the file and follows those instructions instead of its default approach. No code, no dependencies. Just structured instructions.

How do I install a Claude Code plugin?

Run /plugin install irfad7/claude-power-skills inside Claude Code. All 20 skills load immediately and work across every project. You can also copy individual SKILL.md folders into a single project's skills/ directory.

Can Claude Code remember things between sessions?

Not by default. The kairos-session, autodream and session-review skills add that: a structured boot sequence that loads prior context, memory consolidation at shutdown, and cross-session synthesis that merges observations and promotes recurring patterns into reflexes.

How do I stop Claude Code running out of context?

The context-compression skill implements a three-layer pipeline: a roughly 30-40% reduction with no information loss, a 60-70% pass, and a full 85-90% compact for when a conversation has to survive at any cost.

Do these skills work with any model?

They are written for Claude Code specifically, since they use its skill system, slash commands and subagent model. The underlying patterns are portable but the files target Claude Code.

Who made Claude Power Skills?

Irfad Imtiaz, Director of Technology at My Legal Academy and Co-Founder and CTO of Ranql.

Who built Claude Power Skills

Irfad Imtiaz built and maintains Claude Power Skills. He is Director of Technology at My Legal Academy and Co-Founder and CTO of Ranql. These skills came out of running Claude Code on production work every day.

Find him at irfad.me, LinkedIn and GitHub.

More open-source kits by Irfad Imtiaz