ModelSense is an intelligent on-demand advisor that identifies the ideal LLM and effort level for any task by analyzing benchmarks and provider availability.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install modelsense
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install modelsense using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is ModelSense?
ModelSense serves as a sophisticated decision-support tool within the Openclaw Skills ecosystem, designed to remove the guesswork from LLM selection. By acting as an on-demand technical advisor, it helps developers and researchers choose the right model for the right job based on actual benchmark data, task analysis, and cost-efficiency.
Rather than routing requests blindly, ModelSense evaluates the specific dimensions of a task—such as domain, complexity, and required context length—to provide a grounded recommendation. As a core component of Openclaw Skills, it ensures that users leverage their configured providers optimally, whether they need high-speed summarization or deep, research-grade reasoning.
ModelSense Use Cases
- Selecting the most effective model for complex coding tasks like Solidity audits or debugging.
- Identifying cost-efficient flash models for high-volume dialogue and thread summarization.
- Determining the necessary effort level and thinking depth for advanced mathematical conjectures.
- Choosing between different provider tiers like Claude Opus vs. GPT-4o for specific research workflows.
How ModelSense Works
- Analyze the input task to classify it by domain, complexity, and required output format.
- Cross-reference the task requirements against an internal benchmark matrix including HumanEval, GPQA, and MMLU.
- Verify available models by querying the user's active providers through the Openclaw Skills execution tool.
- Map the task to an effort-versus-quality matrix to determine the optimal model tier.
- Generate a structured recommendation including reasoning, cost estimates, and alternative options.
ModelSense Setup
To get started with ModelSense, ensure your environment is configured with your preferred LLM providers. Use the following commands within the Openclaw Skills framework:
# List available models and providers
openclaw models list
# Access the benchmark data configuration
cat data/benchmarks.yaml
ModelSense Data Schema & Taxonomy
ModelSense organizes its recommendation logic through structured data files that are part of the Openclaw Skills directory structure:
| File | Purpose |
|---|---|
data/benchmarks.yaml |
Stores score leaders and task mappings for benchmarks like SWE-bench and GSM8K. |
data/models.yaml |
A frequently updated catalog of model capabilities, context windows, and pricing. |
session_status |
Metadata used to track and apply model switches to the current active session. |
ModelSense Advanced Features
- Auto-Switching: Automatically apply recommended model settings to the active session with a simple confirmation.
- Task Delegation: Spawn dedicated sub-sessions using the best-fit model for specific high-complexity sub-tasks.
- Provider Filtering: Intelligently restricts recommendations to models currently accessible via your Openclaw Skills configuration.
- Benchmark Grounding: Provides technical rationale for every recommendation based on industry-standard datasets.
SKILL.md
Loading
Related Openclaw Skills
Cloudflare DNS Updater
A specialized tool for programmatically creating or updating proxied Cloudflare DNS A records.
Exa AI Search & Research
A comprehensive search and research skill that enables AI agents to perform neural web searches, content extraction, and structured async research.
Find Products
A powerful search and analysis tool for discovering trending ProductHunt products with structured metadata and sentiment insights.
Confirm Form
A specialized utility for generating structured HTML confirmation forms to collect organized feedback and decision-making data from users.
QFPay Payment Integration (HK)
A comprehensive Openclaw Skills integration for the QFPay API, enabling secure multi-method payments for Hong Kong and overseas merchants.
Exa Search
A specialized search integration that leverages the Exa API to deliver structured, LLM-ready web data directly to your environment.
Featured*
