{"version":"1.0","type":"card","id":"acf84178-b67d-4a52-8c4e-6a9b772a4124","url":"https://stacklist.com/card/acf84178-b67d-4a52-8c4e-6a9b772a4124","title":"headroom: Compress Tool for LLM Outputs and Logs","source_url":"https://github.com/chopratejas/headroom","note":"Headroom is a tool designed to compress outputs, logs, files, and RAG chunks before they are processed by large language models (LLMs). It achieves a reduction of 60-95% in token usage while maintaining the same answers, and includes functionalities as a library, proxy, and MCP server.","image":{"url":"https://ucarecdn.com/a2f951e2-3067-467b-942e-3bf6b580d488/","alt":"headroom: Compress Tool for LLM Outputs and Logs","width":1200,"height":600},"stack":{"id":"c691055b-de15-47a1-ad6a-6c4ecec55f18","title":"GitHub repos to check out","url":"https://stacklist.com/stack/c691055b-de15-47a1-ad6a-6c4ecec55f18"},"created_at":"2026-06-18T02:20:33.061Z","updated_at":null,"aco":{"summary":"Headroom is a lightweight JavaScript/Python library that compresses AI agent inputs and outputs by 60-95% through intelligent content routing and multiple compression algorithms. It operates as a library, proxy, or MCP server while maintaining accuracy and reversibility, enabling cost-effective LLM interactions across multiple platforms.","tags":["token-compression","javascript-library","ai-agents","llm-optimization","proxy-server","code-compression"],"key_entities":[{"name":"Headroom","type":"technology","confidence":0.99},{"name":"Claude","type":"technology","confidence":0.95},{"name":"Cursor","type":"technology","confidence":0.9},{"name":"Codex","type":"technology","confidence":0.9},{"name":"LangChain","type":"technology","confidence":0.85},{"name":"Anthropic","type":"organization","confidence":0.9},{"name":"OpenAI","type":"organization","confidence":0.9},{"name":"token-compression","type":"concept","confidence":0.95},{"name":"reversible-compression","type":"concept","confidence":0.9},{"name":"context-compression","type":"concept","confidence":0.95}],"classification":"reference","language":"en","confidence":0.85,"provenance":{"model":"claude-haiku-4-5","tool":"@stacklist/be@0.1.0","confidence":0.85,"timestamp":"2026-06-18T02:20:43.446Z"},"token_counts":{"approximate":3536,"cl100k":3403},"content_hash":"sha256:cbf97f9e16c3331be22eb0953a97440d402729b346473cd321321ef5679cfdac","acp_version":"0.2","body_available":true,"body_tokens":3536,"visibility":"public","agent_accessible":true,"status":"final"},"_links":{"self":"/api/public/card/acf84178-b67d-4a52-8c4e-6a9b772a4124.json","html":"https://stacklist.com/card/acf84178-b67d-4a52-8c4e-6a9b772a4124","md":"/api/public/card/acf84178-b67d-4a52-8c4e-6a9b772a4124.md","stack_json":"/api/public/stack/c691055b-de15-47a1-ad6a-6c4ecec55f18.json"}}