I built token-trim because I kept noticing how much unnecessary text can end up being sent to AI coding assistants — comments, docstrings, blank lines, and formatting that the AI often doesn't need. Since LLMs process text as tokens, sending more unnecessary content means using more of your available context and, with paid APIs, potentially spending more. So I built a small Python CLI that cleans up code and logs before sending them to an AI. It started as a simple experiment, but evolved into a proper open-source, non-profit tool with token counting, savings reports, clipboard support, and directory processing. The goal is simple: help developers spend their AI tokens on the code that actually matters. Would love to hear what you think, and what you'd add next.

token-trim — Save AI Tokens

Trim unnecessary code, save AI tokens, and reduce API costs.

Trim unnecessary code, save AI tokens, and reduce API costs.