[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:headroom":3},"\u003Cdiv align=\"center\">\u003Cpre>\n  ██╗  ██╗███████╗ █████╗ ██████╗ ██████╗  ██████╗  ██████╗ ███╗   ███╗\n  ██║  ██║██╔════╝██╔══██╗██╔══██╗██╔══██╗██╔═══██╗██╔═══██╗████╗ ████║\n  ███████║█████╗  ███████║██║  ██║██████╔╝██║   ██║██║   ██║██╔████╔██║\n  ██╔══██║██╔══╝  ██╔══██║██║  ██║██╔══██╗██║   ██║██║   ██║██║╚██╔╝██║\n  ██║  ██║███████╗██║  ██║██████╔╝██║  ██║╚██████╔╝╚██████╔╝██║ ╚═╝ ██║\n  ╚═╝  ╚═╝╚══════╝╚═╝  ╚═╝╚═════╝ ╚═╝  ╚═╝ ╚═════╝  ╚═════╝ ╚═╝     ╚═╝\n              The context compression layer for AI agents\n\u003C\u002Fpre>\u003C\u002Fdiv>\u003Cp align=\"center\">\u003Cstrong>60–95% fewer tokens (for JSON data), 15-20% fewer tokens (for coding agents) · library · proxy · MCP · content-aware compressors · local-first · reversible\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fchopratejas\u002Fheadroom\u002Factions\u002Fworkflows\u002Fci.yml\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fchopratejas\u002Fheadroom\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg\" alt=\"CI\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fapp.codecov.io\u002Fgh\u002Fchopratejas\u002Fheadroom\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fcodecov.io\u002Fgh\u002Fchopratejas\u002Fheadroom\u002Fgraph\u002Fbadge.svg\" alt=\"codecov\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fheadroom-ai\u002F\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fheadroom-ai.svg\" alt=\"PyPI\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fheadroom-ai\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Fheadroom-ai.svg\" alt=\"npm\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fchopratejas\u002Fkompress-v2-base\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fmodel-Kompress--v2--base-yellow.svg\" alt=\"Model: Kompress-v2-base\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fheadroomlabs-ai\u002Fheadroom\u002Fblob\u002FHEAD\u002FLICENSE\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache%202.0-blue.svg\" alt=\"License: Apache 2.0\" \u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fheadroom-docs.vercel.app\u002Fdocs\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdocs-online-blue.svg\" alt=\"Docs\" \u002F>\u003C\u002Fa>\n\u003C\u002Fp>\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fheadroom-docs.vercel.app\u002Fdocs\" rel=\"nofollow ugc noopener\">Docs\u003C\u002Fa> ·\n  \u003Ca href=\"#get-started-60-seconds\" rel=\"nofollow ugc noopener\">Install\u003C\u002Fa> ·\n  \u003Ca href=\"#proof\" rel=\"nofollow ugc noopener\">Proof\u003C\u002Fa> ·\n  \u003Ca href=\"#agent-compatibility-matrix\" rel=\"nofollow ugc noopener\">Agents\u003C\u002Fa> ·\n  \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FyRmaUNpsPJ\" rel=\"nofollow ugc noopener\">Discord\u003C\u002Fa> ·\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fheadroomlabs-ai\u002Fheadroom\u002Fblob\u002FHEAD\u002Fllms.txt\" rel=\"nofollow ugc noopener\">llms.txt\u003C\u002Fa>\n\u003C\u002Fp>\u003Cp align=\"center\">\u003Csub>\n  \u003Cb>AI agents \u002F LLMs:\u003C\u002Fb> read \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fheadroomlabs-ai\u002Fheadroom\u002Fblob\u002FHEAD\u002Fllms.txt\" rel=\"nofollow ugc noopener\">\u003Ccode>\u002Fllms.txt\u003C\u002Fcode>\u003C\u002Fa> here, or fetch \u003Ca href=\"https:\u002F\u002Fheadroom-docs.vercel.app\u002Fllms.txt\" rel=\"nofollow ugc noopener\">the live index\u003C\u002Fa> \u002F \u003Ca href=\"https:\u002F\u002Fheadroom-docs.vercel.app\u002Fllms-full.txt\" rel=\"nofollow ugc noopener\">full docs blob\u003C\u002Fa>.\n\u003C\u002Fsub>\u003C\u002Fp>\u003Chr \u002F>\n\u003Cp align=\"center\">\u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F20881\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Ftrendshift.io\u002Fapi\u002Fbadge\u002Frepositories\u002F20881\" alt=\"chopratejas%2Fheadroom | Trendshift\" width=\"250\" height=\"55\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\u003Cp>Headroom compresses everything your AI agent reads — tool outputs, logs, RAG chunks, files, and conversation history — before it reaches the LLM. Same answers, fraction of the tokens.\u003C\u002Fp>\n\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fheadroomlabs-ai\u002Fheadroom\u002FHEAD\u002FHeadroomDemo-Fast.gif\" alt=\"Headroom in action\" width=\"820\" \u002F>\n  \u003Cbr \u002F>\u003Csub>Live: 10,144 → 1,260 tokens — same FATAL found.\u003C\u002Fsub>\n\u003C\u002Fp>\u003Ch2>What it does\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Cstrong>Library\u003C\u002Fstrong> — \u003Ccode>compress(messages)\u003C\u002Fcode> in Python or TypeScript, inline in any app\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Proxy\u003C\u002Fstrong> — \u003Ccode>headroom proxy --port 8787\u003C\u002Fcode>, zero code changes, any language\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Agent wrap\u003C\u002Fstrong> — \u003Ccode>headroom wrap claude|codex|grok|copilot|cursor|aider|opencode|cline|continue|goose|openhands|openclaw|vibe|omp|zcode\u003C\u002Fcode> in one command; undo with \u003Ccode>headroom unwrap <tool>\u003C\u002Fcode>\u003C\u002Fli>\n\u003Cli>\u003Cstrong>MCP server\u003C\u002Fstrong> — \u003Ccode>headroom_compress\u003C\u002Fcode>, \u003Ccode>headroom_retrieve\u003C\u002Fcode>, \u003Ccode>headroom_stats\u003C\u002Fcode> for any MCP client\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Cross-agent memory\u003C\u002Fstrong> — shared store across Claude, Codex, Gemini, Grok, auto-dedup\u003C\u002Fli>\n\u003Cli>\u003Cstrong>\u003Ccode>headroom learn\u003C\u002Fcode>\u003C\u002Fstrong> — mines failed sessions, writes corrections to \u003Ccode>CLAUDE.local.md\u003C\u002Fcode> (default, gitignored) or \u003Ccode>CLAUDE.md\u003C\u002Fcode> \u002F \u003Ccode>AGENTS.md\u003C\u002Fcode> \u002F \u003Ccode>GEMINI.md\u003C\u002Fcode> \u002F \u003Ccode>GROK.md\u003C\u002Fcode>\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Output token reduction\u003C\u002Fstrong> — trims what the model \u003Cem>writes back\u003C\u002Fem> (not just what you send): drops ceremony\u002Frestated code and skips deep \"thinking\" on routine steps. See \u003Ca href=\"#output-token-reduction-cut-what-the-model-writes-back\" rel=\"nofollow ugc noopener\">Output token reduction\u003C\u002Fa>.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Reversible (CCR)\u003C\u002Fstrong> — originals\u003C\u002Fli>\n\u003C\u002Ful>\n",1784629918768]