Token Budget Guard

What it does

Token Budget Guard is a high-performance utility designed to save developers from unintentional "context burn." It acts as a pre-execution gatekeeper that analyzes planned operations—like repo ingestions, documentation fetches, or large log reads—and provides an immediate 3-line verdict on the most cost-effective way to proceed.

Why use this skill

Large Language Models (LLMs) often lack the self-awareness to tell you when a command will bloat your context or waste expensive tokens. This skill bridges that gap by providing a standardized estimation framework. Instead of blindly running a grep or reading a 5,000-line log file, your agent will pause to evaluate if the data should be processed in the current session, offloaded to a cheap sub-agent, or skipped entirely.

Problem it solves

• Prevents "context bloat" which degrades model reasoning performance.
• Saves money by flagging operations that would ingest hundreds of thousands of unnecessary tokens.
• Helps you decide when to use expensive models (like Opus) vs. specialized sub-tasks.

Output format

The output is strictly limited to three lines (Verdict, Estimate, and Reason), ensuring it never interrupts your development flow or adds more noise to the conversation.