Can I manually choose a specific model profile?

Yes, you can use flags like --glm or --kimi within your command to override the automatic selection logic and force a specific model.

How does it choose between models?

It analyzes your input for keywords indicating a need for deep reasoning, long context, or simple refactoring, then matches those needs to profiles in your ~/.ccs/config.json.

How do I continue a previous delegation session?

You can use the 'ccs:continue' command pattern, which automatically detects your last session and continues the conversation with the same model profile.

What is CCS Delegation?

It is a skill for Claude Code that automatically routes specific, low-complexity coding tasks to alternative AI models to save costs and improve efficiency.

Is it safe for sensitive architectural changes?

The skill is explicitly designed to exclude security-critical code and complex architectural changes, recommending that those tasks remain within the primary Claude environment.

CCS Task Delegation

Name: CCS Task Delegation
Author: jellydn

byjellydn

•

Productividad y Flujo de Trabajo

Automatically delegates deterministic coding tasks like refactoring and testing to cost-optimized AI models via the CCS CLI.

CCS Delegation streamlines the AI development workflow by intelligently offloading routine, well-defined tasks—such as unit test generation, typo fixing, and documentation updates—to specialized, lower-cost models like GLM or Kimi. By analyzing task requirements for reasoning depth and context length, it automatically selects the optimal model profile from your local configuration, allowing the primary Claude instance to focus on complex architectural decisions and security-critical logic while maintaining high productivity and cost-efficiency.

Características Principales

01Real-time reporting of delegation costs, duration, and execution results

02Seamless session continuation for iterative task completion and follow-ups

03Automatic prompt enhancement with relevant file context and success criteria

0439 GitHub stars

05Support for manual model overrides using simple CLI-style flags

06Intelligent profile selection based on task reasoning and context requirements

Casos de Uso

01Performing repository-wide documentation audits and typo fixes without exhausting primary model limits

02Analyzing large codebases for specific patterns using specialized long-context models

03Generating comprehensive unit tests for new services using cost-optimized models

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jellydn/my-ai-tools ccs-delegation

For use in Claude.ai and ChatGPT

Download Skill