skills.rest
sky770825
sky770825COMMUNITY
Explore all skills by this author →
GitHub

llm-supervisor

Automates cloud-to-local LLM switching with rate-limit detection for OpenClaw workflows.

Updated Feb 25, 2026
Search Tags:#llm#code-generation#ollama#rate-limit#openclaw#agent-workflow#local-fallback
One-click install
npx skills add https://github.com/sky770825/NEUXA- --skill llm-supervisor
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: llm-supervisor
Source: https://github.com/sky770825/NEUXA-/tree/main/quarantine/skills-archive-20260208/llm-supervisor
Command: npx skills add https://github.com/sky770825/NEUXA- --skill llm-supervisor

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Cloud LLM rate limits cause unexpected workflow stops for OpenClaw users, while unvetted switches to lower-quality local LLMs risk producing subpar or broken code for development tasks.

Core Features & Use Cases

  • Automatic Rate Limit Response: Detects cloud LLM rate limit and overload errors in real time, eliminating silent failures or endless retries.
  • Safe Local Fallback: Switches to a local Ollama model when rate limits occur, with mandatory user confirmation for all code generation tasks to prevent quality degradation.
  • Manual Mode Control: Provides slash commands to manually switch between cloud and local LLMs, and ensures all new agents inherit the active LLM mode for consistent workflow behavior.
  • Use Case: Development teams using OpenClaw for code generation can avoid downtime from cloud rate limits while protecting production codebases from unintended low-quality local model outputs.

Quick Start

Use the llm-supervisor skill to automatically switch to your local Ollama model when you hit a cloud LLM rate limit, and require explicit confirmation before running any code generation tasks on the local model.

Frequently Asked Questions about llm-supervisor

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I handle cloud LLM rate limits in OpenClaw agent workflows?▼

To handle cloud LLM rate limits in OpenClaw agent workflows, you can use an automatic fallback mechanism that switches to a local Ollama model when rate limit or overload errors are detected, preventing silent failures and endless retries.

How do I safely fallback to a local Ollama model for code generation?▼

To safely fallback to a local Ollama model for code generation, the workflow implements mandatory user confirmation gates before executing any code generation tasks locally, preventing subpar or broken code from entering production codebases.

Can I manually switch between cloud and local LLMs during an agent workflow?▼

Yes, you can manually switch between cloud and local LLMs during an agent workflow by using slash command interfaces, which ensures that all newly created agents inherit the active LLM mode for consistent behavior.

What happens when a cloud LLM overload error stops my code generation tasks?▼

When a cloud LLM overload error stops code generation tasks, the supervisor applies automatic state tracking to catch the error in real time and gracefully transitions the workflow to a local Ollama model.

How do I prevent low-quality local LLM outputs from degrading my production codebase?▼

To prevent low-quality local LLM outputs from degrading your production codebase, the fallback mechanism requires explicit user confirmation before running any code generation tasks on the local Ollama model.

Related Skills

Explore More Software Engineering Skills→
ollieb89ollieb89

ollama-local

Manage local Ollama models and run chat, completions, and embeddings.

Community
Intermediate
breakcafebreakcafe
1

add-ollama-tool

Integrate an Ollama MCP server to list models and generate outputs in NanoClaw.

Official
Advanced
vulcan365vulcan365

fabrcore-agentmonitor

Captures FabrCore agent messages, events, and LLM calls for monitoring and debugging.

Community
Advanced
latestaiagentslatestaiagents
5

llm-rate-limiting

Implement token bucket, sliding window, and concurrency rate limiting for LLM APIs.

Official
Advanced