New story: Cut Claude Code token cost 4×: let your MCP tool take a URL instead of the pasted file
When an agent gives a tool a long piece of text, the model usually types that text out again as the tool's argument.

New story: Cut Claude Code token cost 4×: let your MCP tool take a URL instead of the pasted file
When an agent gives a tool a long piece of text, the model usually types that text out again as the tool's argument.
Spotify built a Claude Code plugin that reroutes bulk file reads to Gemini Flash - cut their token bill 90% on a Java monorepo. Same math I run on GPU pricing: cheapest box for the job. https://engineering.atspotify.com/2026/9/portal-by-spotify-cut-my-claude-code-token-usage-by-90