LLM-Skills

Keep your system sharp. Reviews your setup, captures learnings, audits for rot, recaps your sessions. Run it regularly to keep improving.

Query AI models in parallel via OpenRouter and synthesize findings. Use when the user says "ask [model]", "get [model]'s opinion", "what does [model] think", "multi-model review", "get a second opinion", "council review", "validate this with [model]", or names any supported model alongside content to review. Requires OPENROUTER_API_KEY set in the environment.

Build ANY interactive HTML artifact backed by a Claude Code channel plugin. The user gets a localhost UI where they can pin comments anywhere on the page or on specific data points; each pin push-delivers to the live Claude session as a channel notification; Claude addresses the question (queries MCPs, runs code, etc.) and updates the artifact in place via tool calls. Use this skill whenever the user wants an interactive visualization, dashboard, explorer, annotation tool, design canvas, data tool, or ANY UI they can "poke at and have Claude answer". Trigger on phrases like "interactive artifact", "pin comments", "channel plugin", "Figma-style comments on data", "annotations that Claude responds to", "build me a dashboard with feedback", "dashboard I can chat with", "tool I can leave notes on", "make an artifact that updates when I ask it questions", "interactive thing backed by a channel", or any reference to the fakechat plugin pattern. This is the right skill any time the user wants more than a one-shot an

This skill should be used when the user asks to "design an ETL pipeline", "build data ingestion", "set up data orchestration", "troubleshoot pipeline issues", "optimize data workflows", or mentions ELT, medallion architecture, batch vs streaming, or data transformation patterns.

This skill should be used when the user asks to "write a prompt", "improve my prompt", "fix this prompt", "optimize a prompt", "learn prompting techniques", "get prompt templates", or mentions prompt engineering, prompt quality, or prompt rewriting. Applies 11 foundational techniques from Forward Future's guide.

This skill should be used when the user asks to "design KPIs", "create a KPI system", "build a measurement framework", "develop a balanced scorecard", "define metrics", "prototype a dashboard", "shortlist measures", "build a KPI Tree", "set up management reporting", or mentions ROKS methodology, KPI definition sheets, stakeholder engagement for measurement, or sustaining KPI systems.

This skill should be used when the user asks to "review installed skills", "find duplicates", "detect skill overlaps", "identify skill gaps", "optimize skills", "audit my skills", or "troubleshoot skill conflicts". Supports Gemini, Claude Code, Cursor, Copilot, Windsurf, and custom setups.

This skill should be used when the user asks to "find a skill", "discover plugins", "search for an MCP", "what plugins exist for X", "fill my skill gaps", "improve my setup", or when Claude recognizes it lacks tools for a task. Searches GitHub and marketplaces to suggest installations.

This skill should be used when the user asks to "write DAX measures", "create Power BI calculations", "help with DAX formulas", "write time intelligence", or mentions aggregations, filters, or DAX performance. Ensures correct syntax, optimal performance, and best practices on the first attempt.