$ loading_
Compress outputs and context before LLMs to cut tokens without losing answers.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "headroom" yet — see the docs or source repo.
Compress these RAG retrieval chunks to 20% of their original length without losing key facts, numbers, or citation relationships, and output streamlined context suitable for sending to an LLM.
A highly compressed retrieval context that preserves essential information and reduces downstream prompt token usage.
Compress the following agent execution logs, tool outputs, and intermediate steps into the smallest useful summary, preserving errors, key parameters, final results, and dependencies for continued LLM reasoning.
A concise log summary that keeps the critical details needed for troubleshooting and reasoning.
Read this large document or code file, then compress it aggressively while preserving structure, key conclusions, and important snippets, and output a version optimized for LLM consumption.
A shorter but high-density representation of the file that an LLM can process at lower cost.
Compress code and prompts, redact placeholders, and cut LLM token usage.
Compress MCP tool schemas to cut tokens while preserving semantics deterministically.
Compress and proxy MCP responses to reduce token usage for LLM tool calls.
Compress LLM text and JSON to reduce tokens and lower usage costs.
Route lightweight text tasks to cheaper models and save main-model tokens.
Compresses LLM conversation context while preserving meaning and reducing token usage.