TokenPressureSandbox v1.0

🎯 Overview

TokenPressureSandbox is a lightweight CLI tool that analyzes prompt token counts and maps them against common LLM context limits (e.g., 4k, 8k, 32k).

This release includes:

• ✅ Token counter using OpenAI-compatible tokenizer (tiktoken)
• ✅ Model context pressure visualization (Safe / Approaching / Truncates)
• ✅ CSV output for programmatic analysis
• ✅ Plaintext .report.txt generation for human-readable logs
• ✅ Portable .bat launcher

No setup hell. Just install dependencies, drop your prompt, and run.

---

💻 Usage

pip install -r requirements.txt
python sandbox.py test_prompt.txt --save-report

Or just run:
run_token_sandbox.bat
📁 Files Included
sandbox.py – main script

test_prompt.txt – demo prompt

token_pressure_output.csv – auto-generated results

token_pressure_report.txt – optional summary (with --save-report)

requirements.txt – dependencies

run_token_sandbox.bat – launch from anywhere

README.md, USAGE.txt – documentation

🛠️ Designed to pair with PromptPressure, run locally in LM Studio, or integrate into any prompt QA pipeline.