← Back to Skills Marketplace
272
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install modelsense
Description
ModelSense — The right model for the right job. Recommends the best LLM model and effort level for any task, based on benchmark data, task analysis, and the...
Usage Guidance
What to consider before installing:
- Functionality: This skill is an advisory — it recommends models and effort levels and can optionally switch the agent's session when you explicitly allow it. If you don't want automatic switching or delegation, decline prompts to 'apply' the recommendation.
- GitHub Actions: The repo includes a CI workflow that optionally uses an OPENROUTER_API_KEY (GitHub secret) to auto-update pricing/context; providing that secret will let the action contact openrouter.ai and commit updated data/models.yaml. If you don't want external updates, don't enable that CI secret or disable the workflow.
- Provider access: To determine which models you can use the skill will try to read your configured providers (it suggests running `openclaw models list`). That reveals which providers/models you have configured (not the secret keys themselves). Ensure the agent runtime has the minimal permissions you expect.
- Review the data files: The benchmarks and model catalog are manually authored; if you rely on their exact rankings/costs for billing or routing, periodically verify accuracy.
- If you need higher assurance: ask the maintainer for the skill's origin (source/homepage) and verify the GitHub Actions secret usage and commit history before enabling automatic updates.
Capability Analysis
Type: OpenClaw Skill
Name: modelsense
Version: 0.1.0
ModelSense is a legitimate utility designed to recommend LLM models based on task requirements and benchmark data. The bundle contains standard Python automation (scripts/update-models.py) for fetching pricing data from the OpenRouter API and a GitHub workflow for periodic updates. The instructions in SKILL.md and AGENTS.md are well-scoped to the tool's purpose, utilizing standard OpenClaw session management tools without any evidence of malicious intent, data exfiltration, or harmful prompt injection.
Capability Assessment
Purpose & Capability
Name/description match the included assets: SKILL.md describes recommendation logic and the repo contains benchmark and model catalogs plus a small updater script to refresh pricing/context from OpenRouter. All requested capabilities (checking local providers, recommending models, optionally switching sessions) are consistent with the stated purpose.
Instruction Scope
SKILL.md instructs the agent to check the user's available providers (suggests running `openclaw models list` via exec or reading from context) and to optionally call session_status/sessions_spawn to switch or delegate. These actions are coherent with recommending and applying models but do require the agent environment to expose provider info and permit session switching; they can access provider metadata (not declared secrets) and act autonomously if allowed.
Install Mechanism
There is no install spec for the agent runtime (instruction-only), which is low-risk. The repository includes a GitHub Actions workflow that installs Python deps (requests, pyyaml) in CI to run scripts/update-models.py; this is a standard, limited CI update pattern and not an agent-time install of arbitrary binaries.
Credentials
The skill declares no required env vars for runtime. The updater script and GitHub Action optionally use OPENROUTER_API_KEY (provided via GitHub Secrets) to fetch pricing/context data — this is reasonable for auto-updating but is an external credential unrelated to runtime recommendations. No other credentials or secrets are requested.
Persistence & Privilege
always:false and disable-model-invocation:false (normal). The repo's CI may commit updated data/models.yaml within the skill repo, which is expected for keeping catalogs fresh. The skill does not modify other skills or system-wide settings.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install modelsense - After installation, invoke the skill by name or use
/modelsense - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
- Initial release of ModelSense: an on-demand advisor to recommend the best LLM model and effort level for any user task.
- Offers clear recommendations grounded in benchmark data, user task analysis, and available providers.
- Provides rationale, cost estimates, and alternative options for different effort or quality needs.
- Supports advisory mode, auto-switching session models, and on-demand task delegation.
- Designed for interactive use when users need help choosing between models or understanding benchmark implications.
Metadata
Frequently Asked Questions
What is Modelsense?
ModelSense — The right model for the right job. Recommends the best LLM model and effort level for any task, based on benchmark data, task analysis, and the... It is an AI Agent Skill for Claude Code / OpenClaw, with 272 downloads so far.
How do I install Modelsense?
Run "/install modelsense" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Modelsense free?
Yes, Modelsense is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Modelsense support?
Modelsense is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Modelsense?
It is built and maintained by xinbenlv (@xinbenlv); the current version is v0.1.0.
More Skills