11 free tools that compress prompts, strip code bloat, split agent contexts, and show you exactly where your tokens burn. Works with OpenAI, Claude, Ollama, LM Studio, and any LLM. Everything runs in your browser.
Strip brackets, quotes, and punctuation noise from JSON configs. Keep structure intact. Paste into prompts at a fraction of the token cost.
Reverse conversion. Reconstruct valid JSON from compact TOON notation.
Many agent frameworks use YAML configs. Strip the dashes and quotes for lighter prompts.
Strip filler words, verbose phrasing, and over-polite bloat from system prompts. Same meaning, way fewer tokens.
See exactly what your prompts cost across 12 models. Daily, monthly, yearly — plus estimated savings after optimization.
Color-coded visualization of token density. See which words and sections are eating your budget.
Compare two prompt versions side by side. See exactly what changed and how it affects token count.
Paste Python, JS, or any code. Strip comments, docstrings, type hints, blank lines. Built for Ollama, LM Studio, and local models.
Allocate your context window. Essential for Ollama, LM Studio, llama.cpp with 4K-32K limits.
Recommended temperature, top_p, top_k, repeat_penalty. Works with Ollama, LM Studio, llama.cpp, vLLM.
Paste your full agent context (agent.md, soul, memory, instructions). Tag each section as Always Send, Review Only, or Skip. Generate minimal payloads for cheap model calls and full payloads for flagship review. Built for the sub-agent workflow — cheap models build, flagships review.