← Back to Skills Marketplace
TurboQuant Optimizer
by
akanji-creator
· GitHub ↗
· v1.0.0
· MIT-0
82
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install turboquant-optimizer
Description
Optimizes OpenClaw token usage via multi-level compression, semantic deduplication, and adaptive token budgeting to reduce API costs and memory footprint.
Usage Guidance
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.
Capability Analysis
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.
Capability Tags
Capability Assessment
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.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install turboquant-optimizer - After installation, invoke the skill by name or use
/turboquant-optimizer - Provide required inputs per the skill's parameter spec and get structured output
Version History
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+.
Metadata
Frequently Asked Questions
What is TurboQuant Optimizer?
Optimizes OpenClaw token usage via multi-level compression, semantic deduplication, and adaptive token budgeting to reduce API costs and memory footprint. It is an AI Agent Skill for Claude Code / OpenClaw, with 82 downloads so far.
How do I install TurboQuant Optimizer?
Run "/install turboquant-optimizer" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is TurboQuant Optimizer free?
Yes, TurboQuant Optimizer is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does TurboQuant Optimizer support?
TurboQuant Optimizer is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created TurboQuant Optimizer?
It is built and maintained by akanji-creator (@akanji-creator); the current version is v1.0.0.
More Skills