Claude Skills by weellio

Control the Agent Ops Center dashboard — open/show it, restart the bridge, reset the agent tiles, or check status. Use when the user asks to open/show the agent dashboard, start or stop the visualizer, or clear the agents.

Autonomously execute a multi-step task — plan it, spawn sub-agents per task, and run to completion without asking for confirmation between steps. Use when the user says "build/do X autonomously", "keep going until done", or "don't ask me, just finish it".

Use when the user wants to create a new Claude Code agent, skill, or slash command — "build me an agent that…", "make a skill that…", "create a command for…". Generates correct YAML front matter + body following best practices and writes the file to the right .claude location (one project, several, or global). Pairs with the Gander dashboard's New-component builder.

Use when the user wants to cut Claude Code token cost or asks "why is this session so expensive / what's eating my tokens / trim my CLAUDE.md / reduce context". Audits the always-on context a project re-sends every turn (CLAUDE.md, memory, MCP servers, skills, agents), ranks it by per-turn token cost, and proposes concrete trims — including a leaner CLAUDE.md rewrite. Pairs with the Gander dashboard's per-session cache-hit / context-fill gauges.