Add Powers to
Your AI Stack
Each skill is a complete, installable capability — system prompt, model recommendation, cost estimate, and example outputs. One click to install. Works with Claude, GPT, Gemini, and OpenRouter.
Free · Cheap
Strips AI tells from any text before it ships to humans
Anything AI-generated that ships to real humans needs a humanizer pass — "unlock," "hope you're well," "biggest takeaway" are dead giveaways.
Variable Cost
Breaks any complex AI task into auditable sequential steps
Most prompts fail because you're asking one model to do too much in one shot.
Cheap
Makes AI show its reasoning before committing to an answer
LLMs give better answers when they show their working — just like humans.
Cheap
Locks exact output format and tone with 2–5 examples
When you can't fully describe what you want in words, show the model.
Moderate
Gives AI agents persistent memory across every session
Stateless agents forget everything. Add memory and they become genuinely useful.
Moderate
Connects any AI to your own documents without fine-tuning
LLMs don't know your documents. RAG fixes that without fine-tuning or retraining.
Saves Money
Cuts your AI API bill by 60–80% without touching quality
You don't need GPT-5.4 for a yes/no question. Smart routing pays for itself immediately.
Moderate
One person doing the output of a full content team
Manual content creation doesn't scale. A pipeline does the work of a team — research to draft to published.
One-Time Setup
Builds AI products that behave consistently in production
Most AI products misbehave because the system prompt is a loose list of rules rather than a structured identity.
One-Time Setup
Hardens any AI product against adversarial users
If you're shipping AI products, adversarial users will try to break your prompts. Most system prompts have no real defence.
Cheap
Mines Reddit threads for real audience pain and product ideas
Reddit is the largest unfiltered repository of real human frustrations. Most people don't know how to read it systematically.
Cheap
Mines old chat history into a reusable skill you own permanently
You've already trained Claude your preferences in past conversations — one prompt extracts that hidden context into a permanent skill.
Variable
Routes any request to the right sub-agent automatically — no babysitting
Move beyond single skills to a top-level orchestrator that fans work out to research, coding, UI, and docs sub-agents automatically.
Cheap
Surfaces your highest-value automation candidates in a 10-minute interview
Most people never figure out what to automate first. The flywheel lets AI interview you about your work and surfaces the right candidates — then builds them.
Variable
Turns any AI project into a domain expert by dropping in a folder of focused skills
The hottest production pattern in Claude: focused skills with clear contracts — no overlap, no babysitting, no mega-prompts. Real teams run 50+ skills this way.
Cheap
Builds every other skill for you — describe a workflow and Claude packages it
Skills are Anthropic's killer feature for turning any repeatable workflow into a packaged module Claude auto-invokes. This is the skill that builds all the others.
Free · Cheap
Runs a 3-pass analysis on any decision — structure, contrarian check, and recommendation
Most decisions get made on the first option that sounds reasonable. This skill interviews you, then runs a contrarian agent that hunts for hidden assumptions and second-order effects you missed.
Free · Cheap
Organises your entire day from calendar, inbox, and tasks in under 60 seconds
The morning ritual of checking calendar, email, Slack, and task lists takes 20–25 minutes. A single /today command replaces all of it with one prioritised plan.
Cheap
Compacts any conversation into a handoff doc a fresh agent picks up and continues without missing a beat
When your context window fills or you need a focused sub-agent, you lose continuity. This skill writes the handoff — context, intent, vibe, and which skills to load next. Inspired by Matt Pocock's viral 70K-star skills repo.
Free · Cheap
Converts any YouTube video or transcript into a long-form article structured to get cited by AI search engines
ChatGPT, Perplexity, and Google AI Overview are the new search. If your content isn't structured for them, it doesn't get cited. This skill converts existing content into GEO-optimised articles — the new SEO.
Free · Cheap
Generates 10 hook variations from any script — contrarian, curiosity gap, stat shock, identity callout, and more
Most content dies in the first 3 seconds. Knowing your hook style is one thing — generating 10 variations of it instantly from any script, in your voice, is another.
Free · Cheap
Scans your newsletters and inbox, kills the noise, surfaces only the 5 things worth knowing today
Most newsletter readers summarise everything — including the promotional drivel. This skill applies a noise filter first, then surfaces only what's genuinely new and actionable for your niche.
Free
Every conversation picks up where the last one left off — automatically
Most people re-explain their context to Claude every single session. A 20-minute memory audit sets up five categories of persistent memory so Claude always starts as a fully briefed collaborator.
Cheap
Your Claude Project set up right — the first time — in 15 minutes
Badly structured projects re-explain context every session. This skill runs a 6-question interview and outputs your complete project architecture: CLAUDE.md, context file plan, folder structure, and launch checklist.
Cheap
Every claim verified before you publish — confidence-rated with source guidance
One wrong statistic destroys months of credibility. This skill runs a three-pass review, rates every claim HIGH/MEDIUM/LOW, and returns a structured report with corrections and recommended sources.
Cheap
Onboard Claude to any context in one document — every session starts fully briefed
Without a CLAUDE.md, every project session starts from zero. This skill runs a 5-section interview and generates a complete, ready-to-install CLAUDE.md: identity, context, rules, task library, and quality standard.
Medium
The AI video brief that gets the shot you imagined — not a blurry approximation
Vague prompts produce generic video. This skill is the Creative Director who translates any concept into a 6-part Higgsfield prompt — subject, action, environment, camera, lighting, mood — plus a full 4–6 shot sequence with iteration options.
Cheap
Your exact writing voice extracted into a system prompt Claude can follow
Paste 3–5 writing samples. Get a precision Voice Fingerprint — 7 dimensions of your style extracted and packaged as an installable instruction set. Claude stops sounding like Claude and starts sounding like you.
Cheap
Build any repeatable workflow into a reusable Claude skill in 10 minutes
Skills are the factory behind every AI automation — one skill.md and Claude auto-invokes the workflow every time it's relevant.
Cheap
Rewrites any vague prompt into a high-performance ICC formula prompt instantly
Generic prompts give generic answers. Instructions + Context + Constraints is the fastest fix that works on every model and every task.
Cheap
AI surfaces the 3-7 gaps in your prompt before generating — first drafts become usable drafts
You don't know what context you're missing. This pattern makes the model surface it before it answers, not after.
Variable Cost
Ends context rot in Claude Code by orchestrating fresh sub-agents for each task
Claude Code sessions degrade halfway through every context window. Context engineering with sub-agents fixes it for hours-long projects.
Cheap
Builds a working MCP server for any tool or API — no docs required
If your favourite tool doesn't have an MCP yet, you can build one. Interview-driven output: complete server.py and install config.
Cheap
Stops Claude generating generic 2014-WordPress-style UI — enforces deliberate design before code
The most installed Claude skill on the platform. Bans Inter, Roboto, and Arial as defaults. Forces real design direction before a line of code is written.
Variable Cost
Builds a coordinated 5-agent AI marketing team with non-overlapping specialist roles
One agent doing everything produces mediocre work. Specialist agents with dedicated skills beat the all-in-one approach every time.
Cheap
Turns any AI skill into a business pitch that closes — outcomes first, technology never
Business owners buy 10 hours back per week, not AI workflows. The framing change is what closes deals.
Turns any repeatable workflow into a 16-line Claude skill file
The framework era is over. One markdown file can replace LangChain, RAG pipelines, and agent boilerplate — if you know the format.
Structures your entire AI workspace with folders and markdown — no frameworks needed
30,000 people use this approach. Folders and markdown files outperform multi-agent frameworks for most real work because they're readable, versionable, and Claude understands them natively.
Reads any financial statement and surfaces the numbers that actually matter
Statements come in every month. The model reads the same document 50 different ways depending on how you prompt. This skill reads it the same right way every time.
Finds every discrepancy between two financial datasets and explains why
Numbers not tying is the most time-consuming part of finance work. This skill runs the reconciliation, categorizes variances, and drafts the explanation — in the time it takes to open a spreadsheet.
Tests and documents internal controls in half the time of a manual audit
Controls testing is the most repetitive part of finance and audit work. Same tests, different data, every quarter. This skill runs the test, documents the evidence, and flags exceptions.
Prevents wasted generation credits by validating video prompts before they run
AI video tools charge per generation. Vague prompts produce unusable results. One pre-flight check cuts failed generations by 60–70% and forces you to specify exactly what you want before spending credits.
Builds AI skills worth trusting by proving the workflow before capturing it
Building a skill too early saves your mistakes and repeats them faster every time. The Map→Prove→Capture→Test loop produces skills that actually work — not ones that looked like they worked once.
Builds a personalised AI learning path from your current skill level to your target role
The generic AI skills lists are all the same. A personalised roadmap that starts from where you are and targets where you want to go is 10x more useful than a list someone made for everyone.
More Capabilities Coming
We're building out 49 more skills mined from 92 AI tutorial videos. Creator tools, coding agents, automation pipelines, and research skills — all free, all installable. Subscribe to get notified when new skills drop.
Get Notified
