# Add to your Claude Code skills
git clone https://github.com/AbsolutelySkilled/AbsolutelySkilled
README.md
What is this?
AbsolutelySkilled is a collection of skills that teach AI agents domain-specific knowledge - from framework APIs to marketing strategy. Each skill is a self-contained folder with structured markdown that agents load on demand.
Skills work with any agent that supports the SKILL.md format: Claude Code, Gemini CLI, OpenAI Codex, Cursor, and 40+ more agents.
Flagship Skills
These skills form the backbone of AbsolutelySkilled - install them first and watch your AI agent transform from a code assistant into an autonomous development partner.
Super Brainstorm - Relentless Design Interview
A structured design interrogation that reads your entire codebase before asking a single question - every assumption challenged, every dependency resolved, one decision at a time.
/absolute-brainstorm "Add real-time notifications to our app"
/absolute-brainstorm "Redesign the authentication flow to support SSO"
/absolute-brainstorm "Build a platform with chat, billing, analytics, and admin dashboard"
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
140,903
gemini
marketing
memory
product-management
seo
skills
software
Marks (Recommended) on every option set - with clear rationale, no wishy-washy "it depends"
No fake options - only proposes multiple approaches at genuine design forks. If the answer is obvious, says so
Decomposes large projects - flags multi-system requests and breaks them into sub-projects automatically
Writes a spec to docs/plans/YYYY-MM-DD-<topic>-design.md
Self-reviews via subagent - catches gaps and contradictions before you see it
You choose what's next - writing-plans, absolute-human, direct implementation, or something else
Absolute-Human - AI-Native Development Lifecycle
A complete SDLC replacement for AI agents - decomposes features into dependency-graphed tasks, executes them in parallel waves with TDD verification, and tracks everything on a persistent board.
/absolute-human "Add OAuth2 login with Google and GitHub providers"
/absolute-human "Migrate the REST API to GraphQL"
/absolute-human "Build a real-time chat feature with WebSocket support"
How it works:
Interview - extracts requirements, constraints, and success criteria
Decompose - breaks work into a dependency-graphed DAG of atomic sub-tasks
Discover - explores your codebase and docs in parallel per task
Plan - creates test-first implementation strategies per task
Execute - runs independent tasks simultaneously via sub-agents in parallel waves
Verify - validates every task with TDD, lint, typecheck, and build checks
Converge - self code review, requirements validation, and full project verification
Everything is tracked on a persistent .absolute-human/board.md that survives across sessions - resume where you left off, audit decisions, or hand off to another developer.
A structured, hierarchical knowledge store in ~/.memory/ that gives your AI agent persistent memory across projects, sessions, and tools - automatically loading only what's relevant to the current context.
# Memories are loaded automatically at conversation start
# After complex tasks, your agent proposes what to remember
"Remember that we always use Bun instead of npm"
"What do you know about our auth system?"
"Update my memory about the deployment process"
How it works:
Tag-indexed retrieval - memories are scored against your current context and only the top matches are loaded (no wasted tokens)
100-line ceiling - files stay small and split automatically, keeping context window usage efficient
Wiki-linked graph - memories cross-reference each other with [[wiki-links]] for rich knowledge traversal
Auto-propose learnings - after complex tasks, your agent suggests what to remember for next time
CRUSP lifecycle - Create, Read, Update, Split, Prune - your memory stays fresh and organized
Skill Audit - Security Analysis for AI Agent Skills
A deep, context-aware security scanner for AI agent skills - combines a Python mechanical pre-scan with AI-powered semantic analysis across 6 threat categories to catch prompt injection, data exfiltration, supply chain attacks, and behavioral risks that regex tools miss.
Prompt injection analysis - reads every instruction as an attacker would: direct overrides, persona hijacking, instruction laundering, conditional triggers, multi-step manipulation
Permission and exfiltration audit - flags destructive commands, credential access, webhook data theft, DNS exfiltration, covert channels
Structured verdicts - PASS / FAIL / REVIEW REQUIRED with severity-ranked findings table and per-finding recommendations
Output as a human-readable table report (default) or JSON for CI/tooling integration.
Video Creator - End-to-End Programmatic Video Production
A 7-step orchestrator that takes you from "make me a video about X" to a finished 4K video - using Remotion, ElevenLabs TTS, royalty-free music, and SFX. No video editing knowledge required.
/video-creator "Create a 60-second product demo for my SaaS dashboard"
/video-creator "Make an explainer video about how our API works"
/video-creator "Build a social clip announcing our new feature"
How it works:
Deep interview - up to 30 questions about your product, audience, tone, assets, and visual preferences
Generate script - structured YAML with scenes, timing, narration text, and audio cues
Visual verification - build minimal Remotion compositions, preview in browser, iterate on design
Build full project - complete Remotion compositions with all animations matching the approved script
Add background audio + SFX - source royalty-free music, place sound effects at trigger points
Add narration - ElevenLabs TTS with voice selection (deferred to save costs until visuals are approved)
4K render - final preview, user approval, render at 3840x2160
Supaguard - Synthetic Monitoring from Your Codebase
A production monitoring skill that reads your source code, generates Playwright monitoring scripts, and deploys them as recurring checks via the supaguard CLI - all without committing any test scripts to your repository.
# Point at your app and the agent does the rest
"Set up monitoring for my Next.js app"
"Monitor the login flow on production"
"Create an uptime check for our checkout page"
How it works:
Reads your source code - scans components, routes, data-testids, API endpoints, and forms