Skip to main content
Glama
deenesjakoruzh
Yang1Bai

claw-tsaver

by Yang1Bai
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

claw-tsaver

OpenClawユーザー向けのトークン節約MCPプロキシ。遅延展開（lazy expansion）により、ツール呼び出しのペイロードを90%以上削減します。

なぜ必要なのか

MCPのツール呼び出しは、1回のレスポンスで数千トークンものHTMLやJSONを返すことがよくありますが、モデルが実際に使用するのはそのうちの5%程度です。残りの95%はコンテキストウィンドウを無駄に消費し、コストを増大させます。claw-tsaverはOpenClawとダウンストリームのMCPサーバーの間に配置され、過大なレスポンスをインターセプトして、モデルにはコンパクトなプレビューとオンデマンドのハンドルを提供します。

Related MCP server: TokenSaver MCP

仕組み

sequenceDiagram
    participant U as OpenClaw (Claude)
    participant C as claw-tsaver proxy
    participant F as fetch / puppeteer / etc.
    U->>C: call_tool("fetch", url)
    C->>F: forward call
    F-->>C: 11,507 tokens of HTML
    Note over C: tiktoken count > threshold
    C->>C: store full content in SQLite
    C-->>U: {preview_head, preview_tail, expand_handle}<br/>(only 104 tokens)
    Note over U: model decides if it needs full text
    U->>C: expand_content(handle)
    C-->>U: full 11,507 tokens

実測値

テスト

元のトークン数

返されたトークン数

削減率

Wikipedia「トークン化（データセキュリティ）」の取得

11,507

104

99.1%

OpenClaw + Claude Sonnet 4.6 + mcp-server-fetchにて検証（2026年4月25日）。 生データ: benchmarks/mvp-day1-fetch.jsonl。

クイックスタート

1. 前提条件

uvをインストールしてください（初回のみ設定）:

curl -LsSf https://astral.sh/uv/install.sh | sh

claw-tsaverのインストールは不要です。uvxがオンデマンドで取得して実行します。

2. ダウンストリームMCPサーバーの設定

~/.claw-tsaver/config.jsonを編集します（claw-tsaver-mcpを初回実行するとテンプレートが自動作成されます）:

{
  "downstream_servers": [
    {"name": "fetch", "command": "uvx", "args": ["mcp-server-fetch"]}
  ],
  "compression_threshold_tokens": 500
}

3. OpenClawへの登録

~/.openclaw/openclaw.jsonのトップレベルに以下のブロックを追加します:

"mcp": {
  "servers": {
    "claw-tsaver": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/Yang1Bai/claw-tsaver",
               "claw-tsaver-mcp"]
    }
  }
}

その後、OpenClawゲートウェイを再起動します: openclaw gateway restart

ダッシュボード

オプション：リアルタイムのトークン節約統計を表示するローカルWeb UI。

uvx --from git+https://github.com/Yang1Bai/claw-tsaver claw-tsaver-dashboard

ブラウザで http://localhost:7878 を開いてください。

ロードマップ

  • [x] モジュールA: 遅延展開プロキシ（今回のリリース）

  • [x] モジュールD: ローカルダッシュボード（今回のリリース）

  • [ ] モジュールB: ツールルーティング（ターンごとに必要なMCPのみを自動ロード）

  • [ ] モジュールC: 会話履歴の圧縮（アトミックな事実カード）

ライセンス

MIT — LICENSEファイルを参照してください。

コントリビューション

IssueやPRを歓迎します。

Available Tools

1 tool
expand_contentA

Retrieve the full content of a previously compressed tool result. Pass the expand_handle returned in the compressed envelope.

ParametersJSON Schema
NameRequiredDescriptionDefault
handleYesThe expand_handle from a compressed result.

TDQS

A4.2/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided; description implies a read operation but does not detail any side effects, auth requirements, or handle expiration.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two efficient sentences front-load the purpose with no wasted words.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a single-parameter tool with no output schema, the description sufficiently guides usage, though output format is unstated.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The description adds context that the handle comes from a compressed envelope, complementing the schema's parameter description.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool retrieves the full content of a compressed result, using specific verb and resource.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description instructs to pass the expand_handle from a compressed envelope, implying the usage context, but lacks explicit conditions or exclusions.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 1 tool updatev0.0.1
    • First observedexpand_content
Deploy Server
Shareable, User-Owned Context BackendPeople, applications, and agents read, write, and coordinate together in one context lake. Reach it over MCP, a CLI, a Python SDK, or REST.fulcradynamics.comAd

TDQS

A4.1/5.0

Scored across 1 tool

Disambiguation5/5

Only one tool exists, so there is no potential for confusion between different tools.

Naming Consistency5/5

With a single tool, there is no naming inconsistency to evaluate; the name 'expand_content' is clear and descriptive.

Tool Count2/5

A single tool for expanding compressed results is very thin; the server likely needs additional tools (e.g., for listing handles) to be useful as an MCP server.

Completeness4/5

The tool covers its stated purpose of retrieving full content, but the server lacks any supporting tools for compression or handle management, leaving a minor gap in the overall surface.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • toon-parse-mcp

    ankitpal181
    A
    license
    Not graded
    quality
    D
    maintenance
    An MCP server that helps AI agents reduce token usage by converting data to TOON format and stripping comments and unnecessary whitespace from code files.
    31 PyPI
    MIT

  • TokenSaver MCP

    pozii
    A
    license
    A
    quality
    D
    maintenance
    An MCP server that reduces AI API costs by up to 97% through token measurement, compression, caching, and pruning, all without changing prompts.
    10
    1
    Apache 2.0

  • token-filter-mcp

    VMexicano
    A
    license
    A
    quality
    B
    maintenance
    An MCP server that intelligently filters and compresses tool outputs to reduce context window usage, saving up to 90% of tokens by removing noise such as passing tests and redundant information.
    5
    12 npm
    MIT

  • smart-context-mcp

    Arrayo
    A
    license
    Not graded
    quality
    C
    maintenance
    MCP server that reduces AI agent token usage by up to 90% through intelligent context compression. Enables efficient code exploration, multi-file refactoring, and debugging by providing tools for smart reading, searching, and managing code context.
    4
    MIT