Cursor Rules, AGENTS.md & Agent Skills Directory
Stop rewriting system prompts from scratch. Copy battle-tested AGENTS.md, Cursor rules, CLAUDE.md, and Claude Code skills (SKILL.md) configured for real production architectures.
FastAPI + Python + PostgreSQL (Async SQLAlchemy & Pydantic v2)
Production guidelines for FastAPI, Pydantic v2 validation, Async SQLAlchemy 2.0 (asyncpg), Alembic migrations, and PostgreSQL/Redis.
LangGraph + Python + FastAPI (Agentic AI Workflows)
Production architecture for autonomous multi-agent systems, LangGraph cyclic state graphs, Pydantic v2 structured tool calling, and Chroma/Qdrant vector stores.
Node.js + Express + MongoDB
Production guidelines for Express REST APIs, Mongoose schema design, and MongoDB indexing/transaction discipline.
Frequently Asked Questions on AI Architecture Rules & Skills
Everything you need to know about setting up AGENTS.md, CLAUDE.md, Cursor .mdc, SKILL.md, and context boundaries.
What is the purpose of AGENTS.md and CLAUDE.md files?
AGENTS.md and CLAUDE.md are specialized system prompt guideline files placed in repository roots. They supply AI coding assistants (like Cursor, Claude Code, Windsurf, and AGY) with non-negotiable architectural boundaries, connection pooling rules, framework idioms, and forbidden anti-patterns to prevent hallucinations and maintain codebase consistency.
How do I use these rules in my local development environment?
Copy the generated AGENTS.md into your project root, or drop the .mdc file into .cursor/rules/ for Cursor — both are automatically indexed. Claude Code directly reads CLAUDE.md for project conventions and CLI commands.
What's the difference between AGENTS.md, CLAUDE.md, and the .mdc file?
AGENTS.md is the open, cross-tool standard read by 30+ agents (Codex, Cursor, Copilot, Jules, and more). CLAUDE.md adds Claude Code-specific CLI commands and conventions. The .mdc file is the modern Cursor Rules format with YAML frontmatter (description, globs, alwaysApply) that replaces the legacy single .cursorrules file.
What is the difference between a rule and an agent skill (SKILL.md)?
Rules (AGENTS.md, CLAUDE.md, .mdc) are always-on project instructions loaded from the repository root. An agent skill is a SKILL.md file with YAML frontmatter (name, description) that agents like Claude Code, Codex, and Cursor load on demand when a task matches its description. Every stack on this page ships in both forms, so you can pin the guidelines globally or let the agent pull them in only when it touches that stack.
How do I install one of these as a Claude Code skill?
Open any rule set and copy the SKILL.md cURL command, or run: curl -s --create-dirs -o .claude/skills/<slug>/SKILL.md https://stackitfast.com/rules/<slug>/SKILL.md. Claude Code discovers project skills under .claude/skills/ automatically; the same file works in any Agent Skills-compatible tool.
Why are stack-specific rules necessary over generic prompts?
Generic system prompts fail on crucial framework-specific edge cases—such as disabling prepared statements in Supabase transaction poolers (prepare: false), configuring Redis noeviction policies for BullMQ, or enforcing asyncpg pool pre-ping in FastAPI. Stack-specific rules provide verified, battle-tested solutions to real integration gotchas.
Are these rules based on real open-source architectures?
Yes. Every rule set in STACK IT FAST is derived directly from verified production repositories benchmarked in our open-source directory, including Cal.com, PostHog, Supabase Studio, Sentry, Grafana, and Dub.
Need rules & skills tailored to your custom project constraints?
Take the 2-minute architecture survey to receive custom stack recommendations and instant AI repository scaffolding prompts.