AI Programming Resource Treasure Trove: Curated Collection of Custom Agents, Workflows, and Plugins

This project is maintained by LEON-sci on GitHub with the original title 'AI-coding-resource-collection'. It aims to address the fragmentation issue of AI-assisted programming tools, compiling resources such as custom agents, instructions, skills, hooks, workflows, and plugins that the author has verified in their work, providing a one-stop reference for AI-assisted programming.

With the widespread application of large language models in the code domain, AI-assisted programming tools have proliferated. However, the fragmentation of the tool ecosystem (different prompt formats, instruction syntaxes, plugin systems) leads to high switching costs for developers; excellent community practices are scattered, making systematic acquisition difficult. This project, as a curated collection, was created to address this challenge.

The repository is categorized by type and scenario, including:

• Custom Agents: Configurations optimized for specific tasks/tech stacks (e.g., front-end, database design agents);
• Instruction Sets: Prompt templates for scenarios like code generation and explanation;
• Skills: Reusable functional modules (e.g., Function Calling formats);
• Hooks: Event-triggered automation scripts (e.g., AI review before code submission);
• Workflows: Multi-step AI-assisted process definitions;
• Plugins: IDE/editor extensions.

1. Quick Task Initiation: Use predefined agents/instructions to quickly handle unfamiliar tasks;
2. Standardized Team Practices: Unify instructions/agents to ensure output consistency;
3. Automate Repetitive Work: Automate tasks like CRUD and unit testing via hooks/workflows;
4. Learning & Experimentation: Study prompt/agent design to improve AI interaction skills.

The resources are compatible with mainstream AI programming tool ecosystems, including:

• OpenAI/ChatGPT (custom instructions, GPTs configurations);
• Claude (Projects, Artifacts instructions);
• Cursor (Composer, Chat prompts);
• GitHub Copilot (Chat instruction tips);
• Open-source solutions (Local LLM, Ollama configurations).

The project adopts a curated model; each resource is screened and tested by the author to ensure quality. The update frequency depends on the author's actual needs. The community can fork and customize it, learning its organization and sharing model.

1. Model version updates may affect prompt effectiveness;
2. Tool ecosystem evolution requires synchronous configuration updates;
3. Understand the principles of the resources to avoid blind use in sensitive/critical scenarios.

This project represents the transition of AI programming from exploration to systematic knowledge management. It is recommended that developers build personal resource libraries to accumulate knowledge assets. In the future, there will be more curated collections and standardization efforts to promote AI-assisted programming as a universal skill.