What is the 'Manual Prototype Step' recommended in this skill?

It is a validation step where you test a single representative input in a model interface before writing code to verify format, quality, and feasibility.

When should I activate the project-development skill?

Activate this skill when starting a new AI project, designing batch pipelines, evaluating task-model fit, or estimating costs for LLM-heavy applications.

What is the difference between single and multi-agent architectures here?

Single-agent is preferred for independent batch items, while multi-agent is used for context isolation and parallel exploration of complex subtasks.

How does this skill handle pipeline state management?

It recommends using the file system as a state machine where each processing stage is marked by the existence of specific files, ensuring natural idempotency and easy debugging.

LLM Project Development Methodology

Name: LLM Project Development Methodology
Author: Kalyanikhandare29

byKalyanikhandare29

数据科学与机器学习

Architects and optimizes LLM-powered applications using structured methodologies, pipeline design, and agent-assisted development patterns.

关于

This skill provides a comprehensive framework for building LLM projects, covering everything from initial task-model fit evaluation to complex multi-agent architectures. It emphasizes a 'manual prototype first' approach, utilizes the file system for state management in pipelines, and provides strategies for cost estimation and structured output parsing to ensure reliable, scalable AI systems. It is ideal for developers moving from simple prompts to production-grade agentic workflows.

主要功能

Structured Output Optimization: Frameworks for crafting prompts that yield reliable programmatic parsing and robust error handling.
Task-Model Fit Evaluation: Criteria to determine if a problem is suitable for LLM processing versus traditional code.
File-System State Management: Patterns for using directory structures to track processing state and ensure idempotency.
Staged Pipeline Architecture: Implementation patterns for discrete, cacheable stages (Acquire, Prepare, Process, Parse, Render).
0 GitHub stars
Cost & Scale Estimation: Formulas to calculate token usage, API overhead, and operational costs for batch processing.

使用场景

Deciding between single-agent and multi-agent architectures based on context window and task complexity.
Designing a high-volume batch processing pipeline for large-scale data analysis or content generation.
Refactoring complex, monolithic LLM scripts into modular, maintainable, and cacheable pipeline stages.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kalyanikhandare29/agent-skills-for-context-engineering project-development

For use in Claude.ai and ChatGPT

Download Skill

GitHub

关于

主要功能

Structured Output Optimization: Frameworks for crafting prompts that yield reliable programmatic parsing and robust error handling.
Task-Model Fit Evaluation: Criteria to determine if a problem is suitable for LLM processing versus traditional code.
File-System State Management: Patterns for using directory structures to track processing state and ensure idempotency.
Staged Pipeline Architecture: Implementation patterns for discrete, cacheable stages (Acquire, Prepare, Process, Parse, Render).
0 GitHub stars
Cost & Scale Estimation: Formulas to calculate token usage, API overhead, and operational costs for batch processing.

使用场景

Deciding between single-agent and multi-agent architectures based on context window and task complexity.
Designing a high-volume batch processing pipeline for large-scale data analysis or content generation.
Refactoring complex, monolithic LLM scripts into modular, maintainable, and cacheable pipeline stages.