Meta tags:
description=Headroom is the context optimization layer for LLM applications. Compress tool outputs, DB results, file reads, and RAG results before they reach the model. Same answers, fraction of the tokens.;

Headings (most frequently used words):

compression, sdk, context, optimization, langchain, vercel, ai, introduction, the, layer, for, llm, applications, quick, preview, community, stats, what, gets, compressed, where, headroom, fits, real, world, results, key, features, framework, integrations, nothing, is, lost, next, steps, lossless, ccr, smart, content, detection, cache, image, persistent, memory, failure, learning, multi, agent, metrics, observability, agno, strands, mcp, tools, typescript, quickstart, installation, proxy, server, how, works, on, this, page,

Text of the page (most frequently used words):
headroom (19), model (15), the (13), compress (12), #compression (12), sdk (12), agent (10), #results (9), more (9), llm (8), tokens (8), vercel (8), tool (8), learn (8), and (7), reads (7), messages (7), result (7), framework (6), integrations (6), langchain (6), guide (6), import (6), from (6), mcp (6), code (6), strands (6), context (6), 100 (6), what (5), proxy (5), for (5), tools (5), agno (5), memory (5), your (4), with (4), content (4), cache (4), integration (4), const (4), openai (4), gpt (4), any (4), typescript (4), claude (4), optimization (4), outputs (4), api (4), saved (4), introduction (4), nothing (3), gets (3), compressed (3), community (3), install (3), quickstart (3), works (3), server (3), ccr (3), full (3), reversible (3), withheadroom (3), anthropic (3), client (3), provider (3), observability (3), ctx (3), token (3), image (3), logs (3), text (3), configuration (3), search (3), savings (3), before (3), same (3), error (3), answers (3), log (3), litellm (3), python (3), rag (3), file (3), keeps (3), next (2), steps (2), lost (2), key (2), features (2), real (2), world (2), where (2), fits (2), stats (2), quick (2), preview (2), get (2), fewer (2), how (2), installation (2), retrieve (2), originals (2), but (2), headroom_retrieve (2), one (2), headroomstrandsmodel (2), agents (2), headroomagnomodel (2), headroomchatmodel (2), per (2), cost (2), breakdowns (2), metrics (2), sharedcontext (2), research (2), failure (2), learning (2), user (2), persistent (2), reduction (2), via (2), router (2), automatically (2), selects (2), resize (2), quality (2), tradeoff (2), json (2), diffs (2), zero (2), compresses (2), statistical (2), analysis (2), function (2), responses (2), preserves (2), drops (2), errors (2), noise (2), everything (2), less (2), fraction (2), layer (2), applications (2), architecture (2), reference (2), this, page, running, minutes, send, goes, into, store, can, fetch, when, needs, detail, aggressive, all, patterns, liner, headroommiddleware, npm, middleware, wrappers, three, cursor, headroom_compress, headroom_stats, wrapping, output, hook, hooks, chatopenai, wrap, chat, supports, retrievers, streaming, async, prometheus, endpoint, request, logging, tracking, budget, limits, pipeline, timing, put, big_output, summary, moves, between, multi, past, sessions, finds, failed, calls, correlates, succeeded, writes, learnings, hierarchical, session, turn, sqlite, hnsw, backends, survives, across, conversations, trained, stabilizes, prefixes, caches, hit, tracks, frozen, preserve, read, discount, auto, detects, html, routes, each, best, compressor, needed, smart, detection, aggressively, stores, gives, details, thrown, away, lossless, 761, 174, github, issue, triage, 254, 502, codebase, exploration, 118, 694, sre, incident, debugging, 408, 765, after, scenario, answer, fatal, was, preserved, not, keyword, matching, field, variance, correct, 260, 144, input, baseline, metric, production, entries, critical, buried, position, changes, transparent, google, bedrock, library, app, optimal, images, change, hunks, unchanged, git, modernbert, classification, removes, redundancy, plain, ranks, relevance, top, matches, failures, passing, build, test, ast, aware, signatures, collapses, bodies, source, anomalies, boundaries, arrays, typical, happens, type, view, detailed, charts, active, instances, 889, requests, optimized, 176, compression_ratio, tokens_saved, print, create, response, tofixed, compressionratio, tokenssaved, console, await, analyze, these, role, database, retrievals, reaches, sees, responds, faster, costs, providers, algorithms, accuracy, open, copy, markdown, they, reach, troubleshooting, handling, help, limitations, benchmarks, simulation, monitoring, management, smartcrusher, getting, started,

Text of the page (random words):
introduction headroom headroom headroom search k getting started introduction quickstart installation community savings compression how compression works smartcrusher code compression image compression text log compression reversible compression reversible compression ccr cache context cache optimization context management memory persistent memory sharedcontext failure learning proxy server proxy server integrations vercel ai sdk openai sdk anthropic sdk langchain agno strands litellm mcp tools configuration configuration observability metrics monitoring simulation api reference api reference architecture architecture benchmarks limitations help error handling troubleshooting introduction introduction headroom is the context optimization layer for llm applications compress tool outputs db results file reads and rag results before they reach the model same answers fraction of the tokens copy markdown open the context optimization layer for llm applications compress everything your ai agent reads same answers fraction of the tokens 87 token reduction 100 accuracy 6 algorithms 100 providers headroom compresses everything your ai agent reads tool outputs database results file reads rag retrievals api responses before it reaches the llm the model sees less noise responds faster and costs less quick preview typescript python import compress from headroom ai const messages role user as const content analyze these results const result await compress messages model gpt 4o console log saved result tokenssaved tokens result compressionratio 100 tofixed 0 from headroom import compress result compress messages model gpt 4o response client messages create model gpt 4o messages result messages print f saved result tokens_saved tokens result compression_ratio 0 community stats 41 8b tokens saved 176 6k cost saved 1 2m requests optimized 889 active instances view detailed charts and breakdowns what gets compressed content type what happens typical savings json arrays tool outputs statistical analysis keeps errors anomalies boundaries 70 90 source code ast aware compression preserves signatures collapses bodies 40 70 build test logs keeps failures and errors drops passing noise 80 95 search results ranks by relevance keeps top matches 60 80 plain text modernbert token classification removes redundancy 30 50 git diffs preserves change hunks drops unchanged context 40 60 images ml router selects optimal resize quality tradeoff 40 90 where headroom fits your agent app tool outputs logs db reads rag results file reads api responses v headroom proxy python library ts sdk or framework integration v llm provider openai anthropic google bedrock 100 via litellm headroom works as a transparent proxy zero code changes a python function compress a typescript function compress or a framework integration langchain agno strands litellm vercel ai sdk mcp real world results 100 production log entries one critical error buried at position 67 metric baseline headroom input tokens 10 144 1 260 correct answers 4 4 4 4 87 6 fewer tokens same answer the fatal error was automatically preserved not by keyword matching but by statistical analysis of field variance scenario before after savings code search 100 results 17 765 1 408 92 sre incident debugging 65 694 5 118 92 codebase exploration 78 502 41 254 47 github issue triage 54 174 14 761 73 key features lossless compression ccr compresses aggressively stores originals gives the llm a tool to retrieve full details nothing is thrown away learn more smart content detection auto detects json code logs text diffs html routes each to the best compressor zero configuration needed learn more cache optimization stabilizes prefixes so provider kv caches hit tracks frozen messages to preserve the 90 read discount learn more image compression 40 90 token reduction via trained ml router automatically selects resize quality tradeoff per image learn more persistent memory hierarchical memory user session agent turn with sqlite hnsw backends survives across conversations learn more failure learning reads past sessions finds failed tool calls correlates with what succeeded writes learnings to claude md learn more multi agent context compress what moves between agents any framework ctx sharedcontext ctx put research big_output summary ctx get research learn more metrics observability prometheus endpoint per request logging cost tracking budget limits pipeline timing breakdowns learn more framework integrations langchain wrap any chat model supports memory retrievers tools streaming async from headroom integrations langchain import headroomchatmodel llm headroomchatmodel chatopenai langchain guide agno full agent framework integration with observability hooks from headroom integrations agno import headroomagnomodel model headroomagnomodel claude agent agent model model agno guide strands model wrapping tool output hook provider for strands agents from headroom integrations strands import headroomstrandsmodel model headroomstrandsmodel agent agent model model strands guide mcp tools three tools for claude code cursor or any mcp client headroom_compress headroom_retrieve headroom_stats headroom mcp install claude mcp tools guide typescript sdk compress vercel ai sdk middleware openai and anthropic client wrappers npm install headroom ai typescript sdk guide vercel ai sdk one liner withheadroom or headroommiddleware for any vercel ai sdk model import withheadroom from headroom ai vercel ai const model withheadroom openai gpt 4o vercel ai sdk guide all integration patterns nothing is lost compressed content goes into the ccr store compress cache retrieve the llm gets a headroom_retrieve tool and can fetch full originals when it needs more detail compression is aggressive but reversible next steps quickstart installation proxy server vercel ai sdk langchain how compression works quickstart get headroom running in 5 minutes install compress and send to your llm with fewer tokens on this page quick preview community stats what gets compressed where headroom fits real world results key features framework integrations nothing is lost next steps