All authors
Claude Skills by KushalP1
github.com/KushalP17 skills0 installs0 views
- AutopilotTurn context-doctor's autopilot on, off, or report its status. Autopilot clears stale tool output from every new Claude Code session's requests, only when the prompt cache is cold, so it never costs more. Use when the user asks to enable, disable, pause or check autopilot.Votes: 0GitHub stars: 2
- CheckupProfile this Claude Code session's context with context-doctor and apply the top fix. Use when the user asks what is eating the context, why the session is slow or expensive, or for a context checkup.Votes: 0GitHub stars: 2
- Compact WindowCompare Claude Code auto-compact window sizes on the user's own recent sessions, and set one if they choose. Use when the user asks about compaction, long or expensive sessions, why sessions get so large, or how to spend fewer tokens automatically (including in the desktop app, where autopilot cannot reach).Votes: 0GitHub stars: 2
- OverheadMeasure what every request re-reads before the user's message (system prompt, tools, MCP schemas, CLAUDE.md, rules, auto memory) and price each memory file per month; offer to split a heavy one. Use when the user asks why sessions start large, what CLAUDE.md or MCP servers cost, or how to trim memory files.Votes: 0GitHub stars: 2
- PackAnswer a question from a large document, log, PDF or folder by reading only the chunks it needs, within a token budget. Use when the user asks about a big file or folder you would otherwise read whole, and you are not going to edit it.Votes: 0GitHub stars: 2
- SavingsShow what context-doctor's levers (compacting at cold resumes, autopilot, an earlier auto-compact window) save, or would have saved, on the user's own recent Claude Code sessions. Use when the user asks how much context-doctor saves, what they spend on input tokens, or whether autopilot is worth turning on.Votes: 0GitHub stars: 2
- Context DoctorKeep the LLM context window lean and fast. Use when the conversation grows long, large content is pasted or produced by tools, the user asks "what's eating my context/tokens", asks to reduce cost or latency of LLM calls, or asks to profile/optimize/compact a conversation. Also triggers on "context doctor", "token usage", "context window", "prompt caching".Votes: 0GitHub stars: 2