Prompts, techniques, and resources for working effectively with LLM coding assistants.
Writing .cursorrules files and project-specific AI instructions for Cursor. Customize LLM behavior per-project.
Extend GitHub Copilot with third-party extensions and custom agents. Build domain-specific LLM capabilities.
Extend Claude Code with MCP servers, hooks, and slash commands. Build custom LLM-powered dev workflows.
Effective system prompts for LLM coding agents. Structure, constraints, and output formatting for reliable code generation.
Prompts for LLM-powered code review. Catch bugs, security issues, and style problems with structured review prompts.
Structured prompts for debugging with LLMs. Reproduce, isolate, and fix bugs systematically using AI assistants.
Prompts for LLM-assisted software architecture design. System design, API design, and database schema generation.
Prompts for generating tests with LLMs. Unit tests, integration tests, and edge case coverage from code context.
Prompts for LLM-assisted refactoring. Extract functions, simplify logic, and improve code structure safely.
Prompts for generating documentation with LLMs. README files, API docs, and inline comments from source code.
Prompts for generating commit messages with LLMs. Conventional commits, meaningful descriptions from diffs.
Use chain-of-thought prompting to improve LLM code generation. Break complex problems into reasoning steps.
Manage LLM context windows for AI coding — token budgets, file pruning, summarization, and retrieval patterns that keep Claude, GPT-4, and Gemini focused on the task.
Design effective workflows for autonomous LLM coding agents. Planning, execution, and verification patterns.
Build everything around the LLM — tools, MCP, hooks, evaluators, memory, sandboxes — so an agent ships code instead of just generating it.
Dan Shapiro's five-level autonomy ladder for AI coding, from spicy autocomplete to the dark software factory — modeled on NHTSA's driving levels.
Techniques for LLM-assisted changes across multiple files. Coordinate edits, maintain consistency, and avoid drift.
Combine TDD methodology with LLM assistants. Write tests first, then use AI to implement passing code.
Twelve community-maintained GitHub awesome lists for AI/LLM coding in 2026 — tools, prompt collections, MCP servers, frameworks, and learning resources, with named maintainers.
Active communities for LLM-assisted development. Discord servers, subreddits, and forums for sharing prompts and workflows.
Compare 2026 coding LLMs: Claude Opus 4.7, Sonnet 4.6, Haiku 4.5, GPT-5, Gemini 2.5, DeepSeek V3.1. Pricing, SWE-bench, and use cases.
Essential LLM prompt engineering techniques for software developers. Get better code output from any AI model.
Best practices for working with LLM coding assistants. Review, test, and integrate AI-generated code safely.
Top VS Code extensions for LLM-assisted development. Code completion, chat, testing, and documentation tools.
Side-by-side comparison: Cursor vs GitHub Copilot vs Claude Code vs Windsurf. Features, pricing, and LLM models.
Using Claude, OpenAI, and Gemini APIs in code. Practical guide to integrating LLM APIs into developer tools and applications.