The original title is "Three open-source approaches to reducing token costs in AI coding agents"

Short summary

A brief post highlighting three open-source repos—rtk, caveman, and superpowers—that tackle token efficiency for AI coding agents from different angles: shrinking input, shrinking output, and avoiding redundant work. The author frames token cost as the new cloud bill for engineers in 2026. Content is mostly links with light commentary.

• •Token cost is the new battleground for AI coding agents
• •Three repos attack the problem: rtk (input), caveman (output), superpowers (dedup)
• •Post is link-heavy with minimal analysis

Generated with AI, which can make mistakes.