← 返回 Skills 市场
TurboQuant Optimizer
作者
akanji-creator
· GitHub ↗
· v1.0.0
· MIT-0
82
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install turboquant-optimizer
功能描述
Optimizes OpenClaw token usage via multi-level compression, semantic deduplication, and adaptive token budgeting to reduce API costs and memory footprint.
安全使用建议
This skill appears to be what it claims: a conversation optimizer that reads sessions, compresses context, and caches results. However, before installing or enabling it on production data: 1) Audit where checkpoints/caches are stored (search for cache paths) and confirm retention, encryption, and deletion behavior. 2) Run it in a sandbox or dev agent to observe filesystem and network activity (ensure it does not exfiltrate data). 3) Review the omitted/truncated library files if possible (the package includes several large source files; ensure no hidden networking or credential-reading code). 4) Because it hooks into pre-api-call, consider enabling it only for specific agents or disabling automatic invocation until you’re comfortable. 5) Verify the repository and maintainer contact (MincoSoft links are provided) and run the included tests/benchmarks (npm test, benchmark) locally. These steps will reduce risk from accidental exposure of chat contents or unexpected persistent storage.
功能分析
Type: OpenClaw Skill
Name: turboquant-optimizer
Version: 1.0.0
The turboquant-optimizer skill is a legitimate utility for OpenClaw designed to reduce token consumption in long conversations through context summarization and deduplication. The core logic in lib/turboquant-optimizer.js and lib/token-budget-manager.js implements text-processing strategies (branded as research-inspired 'Two-Stage Compression') to optimize message arrays before API calls. The CLI tool in bin/turboquant.js reads local session data from the standard OpenClaw directory to provide analysis and benchmarks. No evidence of data exfiltration, malicious command execution, or harmful prompt injection was found; the code's behavior is entirely consistent with its stated purpose of improving efficiency and reducing API costs.
能力标签
能力评估
Purpose & Capability
The name, description, SKILL.md, package.json hooks, CLI and library files all align: this is a token/context optimizer meant to run inside OpenClaw and act on conversation sessions. Minor inconsistency: the registry metadata claimed no required config paths or credentials, but SKILL.md and the CLI/lib code expect/use ~/.openclaw/openclaw.json and read sessions from ~/.openclaw/agents/main/sessions — the skill will therefore access user session files even though 'required config paths' was listed as none.
Instruction Scope
Runtime instructions explicitly state the skill 'monitors all API calls and optimizes context' and the CLI/lib implement reading/parsing session files, deduplication, checkpointing and caching. All of these actions are within the stated purpose, but they give the skill broad access to conversation contents and tool outputs (sensitive data). There are no vague instructions granting additional data collection beyond conversation content.
Install Mechanism
No external installer or remote downloads are used in the provided manifest; install is via openclaw skills install or manual git clone/npm install. All code is bundled in the package and dependencies are only devDeps. That is proportionate and low-risk from an install-source perspective.
Credentials
The skill declares no required environment variables or credentials, which matches the code (no external API keys), but it does access user session files, create checkpoints and caches, and will run on pre-api-call hooks. The metadata omitted required config paths even though the skill reads ~/.openclaw/openclaw.json and session files — this mismatch should be noted because the skill will process potentially sensitive user data and may store caches/checkpoints on disk.
Persistence & Privilege
always:false and normal model invocation are used. package.json registers OpenClaw hooks (pre-api-call, post-session-load) which lets the skill act on every API call once enabled — appropriate for this kind of optimizer but means it will be invoked frequently and see all conversation content while enabled. No indication it modifies other skills or system-wide settings.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install turboquant-optimizer - 安装完成后，直接呼叫该 Skill 的名称或使用
/turboquant-optimizer触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
🚀 TurboQuant Optimizer — Slash AI API costs by up to 85%
Bring cutting-edge Google research to your OpenClaw setup. Delivers
84.8% token reduction with only ~10ms overhead.
REAL RESULTS:
• 13,990 → 1,538 tokens (89% saved) on long conversations
• 5,592 → 318 tokens (94% saved) on repetitive tool calls
• $245 → $37 monthly API costs
• 2.3s → 0.8s response latency
KEY FEATURES:
• Two-stage compression (PolarQuant + QJL-inspired)
• Semantic deduplication of messages & tool results
• Adaptive token budgeting by task type
• Smart conversation checkpointing
• Zero code changes — works transparently
Built by MincoSoft Technologies — Miami-based AI automation
experts serving small businesses nationwide.
🌐 https://mincosoft.com
📞 (866) 667-3063
📧 [email protected]
Initial release of TurboQuant Optimizer for OpenClaw.
- Introduces advanced token and memory optimization system leveraging TurboQuant research (up to 99% token savings).
- Features multi-level optimization: context compression, semantic deduplication, and adaptive token budgeting.
- Provides automatic integration, CLI commands, and programmatic API for optimization, benchmarking, and reporting.
- Implements two-stage compression (PolarQuant + QJL), semantic clustering, token budget visualization, and conversation checkpointing.
- Supplies detailed configuration options, performance metrics, and compatibility with all OpenAI-compatible models on OpenClaw 1.0.0+.
元数据
常见问题
TurboQuant Optimizer 是什么？
Optimizes OpenClaw token usage via multi-level compression, semantic deduplication, and adaptive token budgeting to reduce API costs and memory footprint. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 82 次。
如何安装 TurboQuant Optimizer？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install turboquant-optimizer」即可一键安装，无需额外配置。
TurboQuant Optimizer 是免费的吗？
是的，TurboQuant Optimizer 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
TurboQuant Optimizer 支持哪些平台？
TurboQuant Optimizer 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 TurboQuant Optimizer？
由 akanji-creator（@akanji-creator）开发并维护，当前版本 v1.0.0。
推荐 Skills