shaoji-style-writer
A specialized scriptwriting persona (Shaoji Style) that packages deep existential tragedy within a "wholesome, positive" facade. Expert in constructing…
Skills
Find an Agent Skill for your task. Read the guide and copy the install command.
Updated 11-05 00:16
Agents & models
25,653 results
ab-mapping
Use when the user wants to determine whether a problem can be solved with supervised learning, or tries to turn a business problem into an AI problem. Use when the user asks "Can AI do X", "Does this count as a machine learning problem", or "Is our business scenario suitable for AI". Not applicable to: pure data analysis and insight, exploratory problems without clear inputs and outputs, and non-supervised learning scenarios such as clustering.
ai-transformation-playbook
Apply the "Five-Step Guide to AI Transformation — Organizational Change Roadmap" methodology distilled from AI for Everyone. Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan…
iterate-not-perfect
Apply the "Iterate, Don't Perfect — Progressive Improvement of AI Products" methodology distilled from AI for Everyone. Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan…
ml-workflow
Applies the "ML project workflow — Collect → Train → Deploy → Iterate" methodology distilled from AI for Everyone. Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan…
pilot-momentum-flywheel
Apply the "Pilot Flywheel — Seek Success Before Value" methodology distilled from AI for Everyone. Use when the user needs practical guidance, diagnosis, workflow design, decision support, or an action plan…
loop-5plus1-architecture
Design or audit the component architecture of a complete Loop system: 5 major components + 1 spine. Use when the user needs to build a complete (not minimally viable) loop system, or audit the component completeness of an existing system. Not applicable to: designing a minimally viable loop, or automating a single task. Key triggers: "design a complete loop system", "my loop...
loop-three-elements
The starting point for designing or auditing any loop system: break the loop down into the three elements of Trigger, Action, and Stop Condition. Use when the user needs to design a new loop, diagnose a failing loop, or explain to others "what a Loop is." Not applicable to: single task execution, non-loop automation scripts, or systems that already run stably and do not need redesign. Key…
autopilot
Install or refresh a repo-local Claude autopilot bundle into the current repository or an explicitly provided target repository. Use when asked to set up…
model-enhancement-servers
Guide for creating MCP servers that enhance LLM reasoning through structured processes, persistence, and workflow guidance. Use when building MCP servers for…
backward-compatibility-reviewer
Expert agent for reviewing Keboola component PRs with focus on backward compatibility — ensuring existing user configurations, sync actions, and output tables…
skill-writer
Expert guide for creating, structuring, and documenting Agent Skills following the official specification. Use when you need to create a new Agent Skill,…
lead
Orchestrate an ongoing task with AGY while the current agent owns key decisions, review, and delivery. Use when the user invokes /agy:lead or asks you to…
staffer
Delegate a general-purpose task to Google's Antigravity CLI (agy staffer, fast Gemini) with a minimal, unopinionated prompt. Use when the user says…
audit-llm-security
Read-only OWASP LLM Top 10 audit of app-facing AI: prompt injection, data leakage, unsafe output or agency, RAG risks, unbounded spend. Use when "audit LLM…
ralph
Convert PRDs to prd.json format for the Ralph autonomous agent system. Use when you have an existing PRD and need to convert it to Ralph's JSON format.
instagram-dm-sales-agent
Automated Instagram DM outreach agent using Playwright browser automation. Sends personalized cold DMs to leads at scale using the Inbox Compose Method.…
sharpen-saw
Audits past Claude Code session transcripts offline to find recurring agent friction (failing commands, edit spirals, permission denials, context hogs, dormant…
create-ai-tool
Analyze the user's requirements to determine the appropriate type, either Skill or Agent, and delegate its creation. Activated when creating a tool, skill vs agent, what to build, or when a tool type selection is requested.
create-skill
Creates a Claude Code Skill or improves an existing skill. Activated when requested to create a skill, modify a skill, write SKILL.md, make a new skill, improve a skill, or update a skill.
progressive-disclosure
Three-step information loading principle for efficiently using the LLM context window. Reference when writing Skills, Agents, and Prompts. Activate when writing SKILL.md and AGENT.md, designing prompts, or optimizing context.
update-ai-tools
After adding/changing AI tools (Skill, Agent), recursively update all README.md files and reflect them in the CHANGELOG. Use when requested to update-ai-tools, update README, refresh tool docs, or sync documentation.
companion-smoke-guide
Verify the harmless Companion Smoke Test MCP connection and report its status without changing any external state.
find-skills
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express…