tare MCP server
Commit 9 Nov 2026Mengurangi penggunaan context window LLM sebesar 44–73% melalui kompresi yang lossless, cache-safe, dan output-aware untuk coding agent.
Juga dikenal sebagai mstuart/tare
- Stars
- 5
- Tools
- —
- Unduhan, 30 hari
- —
- Versi terbaru
- —
- Commit terakhir
- 9 Nov 2026
- Lisensi
- MIT
Apa itu tare MCP?
Mengurangi penggunaan context window LLM sebesar 44–73% melalui kompresi yang lossless, cache-safe, dan output-aware untuk coding agent.
Ringkasan dibuat dari README proyek
Tentang tare MCP server
Developer yang sedang membangun coding agent dapat mengurangi penggunaan context window sebesar 44% hingga 73% menggunakan tare. Tool ini berada di antara agent dan model API untuk mengompresi tool outputs, logs, dan JSON ke dalam bentuk yang lebih padat. Tool ini dirancang agar lossless secara default, yang berarti informasi hanya akan dihapus jika Anda secara eksplisit memilih metode seperti AST code skeletonization atau telegraphic natural language.
User dapat men-deploy ini sebagai local proxy dengan mengarahkan base URL agent ke tare-proxy, yang mendukung endpoint Anthropic dan OpenAI. Tool ini juga tersedia sebagai CLI dengan subcommand seperti compress dan skeletonize, atau sebagai MCP server untuk client seperti Claude Desktop dan Cursor. Sistem ini bersifat cache-aware, mendeteksi breakpoint untuk memastikan provider prefix caches tetap efektif, dan bersifat output-aware, yang akan mengurangi tingkat kompresi jika mendeteksi lonjakan verbosity pada model.
Integrasi sangat fleksibel melalui tare-mcp server yang tidak memerlukan API key karena beroperasi via stdio, atau melalui library Python dan JavaScript untuk in-process compression.
Apa yang bisa Anda lakukan dengannya
- 1Mengompresi file source yang besarDeveloper dapat menggunakan skeletonization untuk menghapus isi fungsi (function bodies) sambil tetap mempertahankan structural signatures dari sebuah file.
- 2Mengurangi overhead log dan JSONAnda dapat memperkecil data tabular besar atau file log agar lebih banyak informasi dapat masuk ke dalam context window.
- 3Integrasi dengan client MCPUser dapat menambahkan kompresi dan persistent memory ke client MCP mana pun tanpa memerlukan API key.
Ditulis berdasarkan README proyek. Setiap use case diperiksa terhadap kutipan dari README tersebut.
Konfigurasi
TARE_UPSTREAM
| Nama | Deskripsi | Tipe |
|---|---|---|
| TARE_UPSTREAM | — |
Pengecekan otomatis
Berjalan 14 Nov 2026Laporan lengkap →Belum diverifikasi. Sebuah server memerlukan skor 80, tidak ada temuan serius, dan data yang cukup untuk kami periksa. Cara kerja checks · Yang perlu diperbaiki untuk menambah poin
- Source repository terhubung dan dapat dijangkaumstuart/tare is public and reachablegithub.com/mstuart/tare ↗Lolos
- Nama paket mencantumkan repositori ini sebagai sumbernyaTidak diperiksa
- Provenance build (atestasi Sigstore)Tidak diperiksa
- Lisensi open source (OSI)MITspdx.org/licenses/MIT.html ↗Lolos
- Ada commit atau release dalam satu tahun terakhirLast commit or release 2026-11-09github.com/mstuart/tare ↗Lolos
- OpenSSF ScorecardTidak diperiksa
- Nama package tidak mirip dengan package populerTidak diperiksa
- Tidak ada advisory kritis pada paket atau dependensinyaTidak diperiksa
- Tidak ada instruksi tersembunyi untuk model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersLolos
- Tools yang bersifat destruktif telah diberi labelTidak diperiksa
- Endpoint remote merespons handshake MCPTidak diperiksa
README
Lihat di GitHubtare sits between your agent and the model API and shrinks the context window. Three properties make it different:
- Lossless by default — tool output, logs, and JSON are re-encoded into a denser equivalent form. Information is dropped only when you opt in (row-caps, telegraphic NL, AST code skeletonization).
- Cache-correct — provider prefix caches discount cached tokens ~10×, and one rewritten byte in the prefix forfeits it. tare detects the cache breakpoint and compresses only the dynamic suffix.
- Output-aware — over-compression makes models compensate with verbose output, so total cost can rise as input falls. tare watches output tokens per turn and backs off when they spike.
What you get
tare-proxy— point your agent's base URL at it; zero code changes. Speaks Anthropic (/v1/messages) and OpenAI (/v1/chat/completions).tareCLI —compress,skeletonize,compact-lossy, and 12 more subcommands (full reference).tare-mcp— compression + persistent cross-session memory as MCP tools, for any MCP client.- Libraries —
tare-core(Rust),tare-compress(Python),tare-ai(JS/TS).
Proof
All numbers are measured by running target/release/tare on the committed corpus with tiktoken
o200k_base (v0.13.0). Results live in crates/tare-bench/results/; reproduce with
python3 crates/tare-bench/run_proof.py.
| Name | Content type | Command | Input tokens | Output tokens | Reduction |
|---|---|---|---|---|---|
| cargo_packages | json_array | tare compact-lossy |
6,906 | 3,625 | 47.5% |
| ps_aux | tabular | tare compact-lossy |
1,545 | 802 | 48.1% |
| app_log | logs | tare compact-lossy |
13,217 | 6,551 | 50.4% |
| agent_context | agent_context | tare compress |
15,130 | 8,499 | 43.8% |
| server_rs | code | tare skeletonize --path server.rs |
5,930 | 1,582 | 73.3% |
| json_crush_rs | code | tare skeletonize --path json_crush.rs |
3,937 | 1,607 | 59.2% |
| readme_prose | prose | tare compact-lossy |
5,732 | 2,727 | 52.4% |
Code reads are ~67–76% of a coding agent's tokens (SWE-Pruner, ACL 2026),
so skeletonization is the biggest lever. Head-to-head harnesses against Headroom, LLMLingua-2,
lean-ctx, and RTK live in crates/tare-bench/benchmarks/.
Compared to
| Scope | Deploy | Local | Lossless default | Output-aware¹ | |
|---|---|---|---|---|---|
| tare | tools · logs · files · JSON · history | proxy · library · CLI · MCP | ✅ | ✅ | ✅ |
| Headroom | all context | proxy · lib · MCP | ✅ | ❌ (reversible via cache) | ❌ |
| RTK | CLI command outputs | CLI wrapper | ✅ | ❌ | ❌ |
| lean-ctx | CLI commands, MCP tools | CLI · MCP | ✅ | ❌ | ❌ |
| LLMLingua-2 | prose / RAG | library (ML model) | ✅ | ❌ | ❌ |
| OpenAI / Anthropic native compaction | conversation history | provider-native | ❌ | ❌ | ❌ |
¹ Output-aware = reads the model's output token count each turn and reduces compression aggression
when verbosity spikes (the x-tare-verbosity-spike signal in
crates/tare-proxy/src/server.rs).
When to use · When to skip
Great fit if you…
- run coding agents and want savings without losing information by default
- care about the provider cache staying warm (tare won't break the prefix)
- want code reads compressed structurally (signatures kept, bodies elidable on demand)
Skip it if you…
- only use a single provider's native compaction and don't need a cross-provider proxy
- run in a sandbox where a local proxy process can't run
Get started (60 seconds)
# 1 — install (no Rust toolchain needed)
curl -fsSL https://raw.githubusercontent.com/mstuart/tare/main/install.sh | sh # → ~/.local/bin
# or: npm install -g tare-ai
# or: docker pull ghcr.io/mstuart/tare
# or: cargo install tare-cli
# or: git clone https://github.com/mstuart/tare && cd tare && cargo build --release
# 2 — run as a proxy (point your agent's base URL at http://localhost:8787)
TARE_UPSTREAM=https://api.anthropic.com tare-proxy
# 3 — or use the CLI on any stdin
cat big.rs | tare skeletonize --path big.rs # drop fn bodies, keep structure
ps aux | tare compact-lossy --max-rows 30 --max-field 110
| Env var | Default | Meaning |
|---|---|---|
TARE_UPSTREAM |
https://api.anthropic.com |
upstream API base URL |
TARE_PORT |
8787 |
listen port |
TARE_RECENCY |
4 |
recent tool outputs always kept |
TARE_ENABLED |
true |
set 0/false for byte-exact passthrough |
TARE_CONTEXT_LIMIT |
200000 |
model context window (drives the fill dial) |
TARE_OUTPUT_HOLDOUT |
0 |
fraction of sessions left uncompressed (A/B for tare output-savings) |
TARE_ADMIN_TOKEN |
unset | secret required by admin endpoints and CLI admin commands; unset disables the admin API |
TARE_LOG |
unset | set it to log one line per turn with the compression report |
Response headers (x-tare-input-tokens, x-tare-net-tokens, x-tare-dropped, x-tare-aggression,
x-tare-verbosity-spike, x-tare-halted) report what each turn did; GET /admin/stats and
POST /admin/runtime-env expose live stats and hot config when TARE_ADMIN_TOKEN is set; clients
must send the token in x-tare-admin-token. Details: getting started.
Integrations
# pip install tare-compress
import tare
out = tare.compress(blocks_json, task="fix the login bug") # in-process, no proxy needed
// npm install tare-ai
import { withTare } from "tare-ai";
const client = new Anthropic(withTare({ apiKey: "..." })); // routes through the local proxy
Python ships all core transforms in-process plus adapters for litellm, ASGI, LangChain, Agno, and Strands; JS/TS ships proxy helpers for the Anthropic/OpenAI SDKs and the Vercel AI SDK. Full adapter reference: docs/integrations.md.
Use with your Claude subscription — MCP, no API key
The proxy forwards whatever auth the client sends — a billable x-api-key, or your Claude Pro/Max
subscription OAuth token when you point Claude Code's ANTHROPIC_BASE_URL at it
(scripts/live-smoke-sub.sh runs exactly this round-trip). Prefer no base-URL change at all? Run
tare-mcp: a local stdio server your agent calls as tools — it never calls the model itself, so it
needs no API key.
# easiest — no build; npx fetches the prebuilt binary on first run:
claude mcp add tare -s user -- npx -y -p tare-ai tare-mcp
{
"mcpServers": {
"tare": { "command": "npx", "args": ["-y", "-p", "tare-ai", "tare-mcp"] }
}
}
Paste the same block into any MCP client (Cursor, Codex, Claude Desktop's
claude_desktop_config.json, …). It exposes 10 tools: compression (tare_compress,
tare_skeletonize, tare_compact_lossy, tare_deref_images), a reversible tare_expand,
tare_stats, and cross-session memory — full list.
Wrap your agent
One command starts the proxy and launches your CLI agent through it — ENV-based and ephemeral:
tare wrap claude # start proxy + launch Claude Code through it
tare wrap codex --port 9000 # 12 agents supported: claude, codex, aider, goose, …
tare wrap claude --print # dry-run: show what would run
Full agent matrix and modes: docs/cli.md.
Status & limitations
- Tested — verified end-to-end against the live Anthropic API (a full proxy round-trip on a
Claude subscription, plus the MCP server over real stdio JSON-RPC), on top of 228 unit,
integration, and property tests;
fmt/clippy -D warnings/cargo denygate every commit. - Deploy it as a local sidecar — tare runs next to your agent and forwards your credentials
upstream without logging or persisting them; treat it as a trusted component on your own machine,
not shared multi-tenant infrastructure (SECURITY.md). Startup failures exit with a
clear
[tare-proxy] fatal: …message, never a panic backtrace.
Known edges
- Proxy and CLI token counts are approximate (
tare-tokenize, chars/4); the benchmark numbers above are measured with tiktoken. - The context-fill signal counts the serialized request (incl. JSON envelope), so it slightly over-estimates fill — conservative, errs toward compressing sooner.
- A
>2 MBstreaming response whose final usage event straddles the 64 KB tail buffer may skip one verbosity sample (non-fatal).
Non-goals — no trained ML text-compressor (no weights to download, no inference latency) and no
audio: transcribe externally and feed the transcript through tare compress.
Architecture
Nine crates: engine (tare-core), proxy + closed-loop controller (tare-proxy), CLI, MCP server,
SQLite memory, tokenizer, cache models, Python bindings, bench harness. Optional neural-embed
feature swaps keyword relevance for exact-cosine neural embeddings.
Diagram and crate guide: docs/architecture.md.
Contributing
See CONTRIBUTING.md. CI runs cargo fmt --check, cargo clippy -D warnings,
cargo test, and a release build — please make sure those pass locally.
License
MIT — see LICENSE.
Pertanyaan tentang tare MCP
Bagaimana tare menangani kompresi context?
tare menggunakan kompresi lossless-by-default yang melakukan re-encode pada tool outputs, logs, dan JSON ke dalam bentuk padat yang setara, namun tetap cache-correct untuk provider prefix caches.
Bisakah saya menggunakan tare dengan langganan Claude saya tanpa API key?
Ya, dengan menjalankan tare-mcp sebagai local stdio server, agent Anda akan memanggilnya sebagai tools tanpa server perlu memanggil model itu sendiri, sehingga tidak memerlukan API key.
Apakah tare merusak provider prefix caching?
Tidak, tool ini dirancang agar cache-correct dengan mendeteksi cache breakpoint dan hanya mengompresi bagian dynamic suffix saja.