mcp-to-skill

Convert MCP servers into on-demand skills. MCP tool schemas sit in the system prompt on every turn (500-2000 tokens per tool, regardless of whether they're used). Skills cost zero tokens until loaded via view. For a typical setup with 4-5 MCP servers exposing 20-40 tools, this reclaims 10,000-30,000 tokens of context per turn.

This matters because that's 10-30% of the context window burned before the conversation even starts — and it compounds: every turn re-injects the full schema.

Not every MCP should become a skill. Apply this heuristic:

Convert MCP (Model Context Protocol) servers into on-demand skills to dramatically reduce active context window usage. MCP tool schemas consume tokens on every turn whether used or not; skills load only when needed. Analyzes MCP tool definitions, classifies each tool by replacement strategy, and generates a complete skill package. Use this skill whenever the user mentions: convert MCP, MCP to skill, reduce context size, too many tools, tool token bloat, slim down MCPs, context window full, replace MCP, MCP migration. Also trigger when the user has many active MCP servers and wants to optimize context usage, asks about reducing token overhead from tool definitions, or says my context is too big or running out of context when MCPs are active. Source: mathews-tom/praxis-skills.