headroom

---

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.

What it does

• Library — compress(messages) in Python or TypeScript, inline in any app
• Proxy — headroom proxy --port 8787, zero code changes, any language
• Agent wrap — headroom wrap claude|codex|copilot|cursor|aider|opencode|cline|continue|goose|openhands|openclaw|vibe|zcode in one command; undo with headroom unwrap <tool>
• MCP server — headroom_compress, headroom_retrieve, headroom_stats for any MCP client
• Cross-agent memory — shared store across Claude, Codex, Gemini, auto-dedup
• headroom learn — mines failed sessions, writes corrections to CLAUDE.local.md (default, gitignored) or CLAUDE.md / AGENTS.md / GEMINI.md
• Output token reduction — trims what the model writes back (not just what you send): drops ceremony/restated code and skips deep "thinking" on routine steps. See Output token reduction.
• Reversible (CCR) — originals are cached for retrieval on demand

How it works (30 seconds)

 Your agent / app
   (Claude Code, Cursor, Codex, LangChain, Agno, Strands, your own code…)
        │   prompts · tool outputs · logs · RAG results · files
        ▼
    ┌────────────────────────────────────────────────────┐
    │  Headroom   (runs locally — your data stays here)  │
    │  ────────────────────────────────────────────────  │
    │  CacheAligner  →  ContentRouter  →  CCR            │
    │                    ├─ SmartCrusher   (JSON)        │
    │                    ├─ CodeCompressor (AST)         │
    │                    └─ Kompress-v2-base (text, HF)  │
    │                                                    │
    │  Cross-agent memory  ·  headroom learn  ·  MCP     │
    └────────────────────────────────────────────────────┘
        │   compressed prompt  +  retrieval tool
        ▼
 LLM provider  (Anthropic · OpenAI · Bedrock · …)

• ContentRouter — detects content type, selects the right compressor
• SmartCrusher / CodeCompressor / Kompress-v2-base — compress JSON, AST, or prose
• CacheAligner — stabilizes prefixes so provider KV caches actually hit
• CCR — stores originals locally; LLM calls headroom_retrieve if it needs them

→ Architecture · CCR reversible compression · Kompress-v2-base model card

Get started (60 seconds)

# 1 — Install
uv tool install "headroom-ai[all]"      # Install `headroom` CLI as a global tool in self-contained virtual env
pip install "headroom-ai[all]"          # Python — ships the `headroom` CLI
npm install headroom-ai                 # TypeScript SDK only — no `headroom` CLI

# 2 — Pick your mode  (the `headroom` commands below come from the uv or pip install)
headroom wrap claude                    # wrap a coding agent
headroom proxy --port 8787              # drop-in proxy, zero code changes
# or: from headroom import compress      # inline library

# 3 — Verify setup and see the savings
headroom doctor                         # health check — confirms routing is working
headroom perf
headroom dashboard                      # live savings dashboard (proxy must be running)

To use headroom, it is recommended you launch a wrapped agent session each time so that all necessary setup is completed. When wrapping a coding agent, headroom starts a local proxy, sets up an MCP server that provides tools such as rtk and tokensave, and launches a coding agent session configured to proxy requests to headroom.

The headroom CLI ships only via the PyPI package. The npm headroom-ai is the TypeScript SDK — a library you import (import { compress } from 'headroom-ai'), not a CLI, so it provides no headroom command.

Granular extras: [proxy], [mcp], [ml], [code], [memory], [vector] (optional HNSW backend — needs a C++ toolchain, not in [all]), [relevance], [image], [agno], [langchain], [evals], [pytorch-mps] (Apple-GPU memory-embedder offload — set HEADROOM_EMBEDDER_RUNTIME=pytorch_mps). Requires Python 3.10+.

Codex / global install

If Codex or another MCP client cannot inherit a shell PATH reliably, install Headroom as a persistent uv tool and point the client at the absolute binary path:

uv tool install "headroom-ai[all]"
command -v headroom

Then use the returned path in MCP config:

[mcp_servers.headroom]
command = "/absolute/path/from/command-v/headroom"
args = ["mcp", "serve"]

command = "headroom" only works when the client starts with a PATH that already includes the uv tool directory.

Proof

Savings on real agent workloads:

Workload	Before	After	Savings
Code search (100 results)	17,765	1,408	92%
SRE incident debugging	65,694	5,118	92%
GitHub issue triage	54,174	14,761	73%
Codebase exploration	78,502	41,254	47%

Accuracy preserved on standard benchmarks:

Benchmark	Category	N	Baseline	Headroom	Delta
GSM8K	Math	100	0.870	0.870	±0.000
TruthfulQA	Factual	100	0.530	0.560	+0.030
SQuAD v2	QA	100	—	97%	19% compression
BFCL	Tools	100	—	97%	32% compression

Reproduce: python -m headroom.evals suite --tier 1 · Full benchmarks & methodology

Output token reduction (cut what the model writes back)

Everything above shrinks the prompt you send. But you also pay for every token the model writes back — and on Opus-class models output costs 5× input. A lot of that output is waste: "Great, let me…" preambles, re-printing code you just showed it, and deep "thinking" on routine steps like reading a file.

Headroom can trim that too, from the proxy, without you changing any code:

• Verbosity steering — appends a short "be terse, don't restate context" note to the end of the system prompt (so your prompt cache still hits).
• Effort routing — when a turn is just the model resuming after a tool result (a file read, a passing test), it dials the model's thinking effort down. New questions and errors keep full effort.

Applies to Anthropic /v1/messages and OpenAI-compatible endpoints (/v1/chat/completions, /v1/responses). Effort routing uses reasoning_effort on OpenAI, thinking.budget_tokens / output_config.effort on Anthropic — same clamp-only invariant on both paths, same output_shaper:* label vocabulary.

Turn it on:

export HEADROOM_OUTPUT_SHAPER=1     # off by default
headroom proxy --port 8787

Already running a proxy? These switches are read live on every request, so a proxy that headroom wrap reused (rather than started) would not see a value you export afterwards — its environment was snapshotted at launch. headroom wrap now hot-syncs your current settings to the running proxy via a loopback POST /admin/runtime-env, so they take effect immediately with no restart (no cold start, no dropped requests, no lost caches). Set them before you wrap. On a shared proxy these overrides are global — the last explicit setting wins.

Learn the right terseness for you. People don't say how terse they want answers — they show it (they interrupt long replies, or move on before they could have read them). headroom learn --verbosity reads your past sessions and picks the level automatically:

headroom learn --verbosity            # preview what it found (dry run)
headroom learn --verbosity --apply    # save it; the proxy uses it from now on

See how many output tokens you saved. Output savings are counterfactual — we never see what the model would have written — so Headroom reports an honest estimate with a confidence range, never a made-up number:

headroom output-savings
# Reduction: 31.7%  (95% CI 27.7% … 35.7%)   [estimated]

Want a measured number instead of an estimate? Leave 10% of conversations unshaped as a control group: export HEADROOM_OUTPUT_HOLDOUT=0.1. The dashboard shows an Output Tokens Saved card next to input compression, labelled measured or estimated with the confidence band.

→ Full write-up incl. the measurement methodology: Output token reduction

Agent compatibility matrix

| Agent | `headr