Write the AGENTS.md that the loop reads at the start of every run — project context, conventions, and loop architecture in one file
AGENTS.md Template for Loop Projects: Claude Code Prompt
Write the AGENTS.md that the loop reads at the start of every run — project context, conventions, and loop architecture in one file
Write an AGENTS.md file for [YOUR_PROJECT] that is designed to be read by a loop agent at the start of every run — not by a human once and forgotten. The file must include: **Project context** (under 50 words): What th
Agent loop memory prompt for writing the AGENTS.md the loop reads every run — context, conventions, architecture, and escalation rules in under 400 words.
Agent loop memory prompt for writing the AGENTS.md the loop reads every run — context, conventions, architecture, and escalation rules in under 400 words.
Start by clearly defining your project context to ensure the loop agent understands the essential details, including the tech stack and sprint goal, within the 50-word limit.
Use the placeholders effectively to outline the loop architecture, detailing the cadence, roles of sub-agents, and available tool connections to facilitate seamless task execution.
Add specific and measurable criteria in the definition of done section to provide clarity on task completion, ensuring the loop agent knows exactly when to mark tasks as complete.
What is the AGENTS.md Template for Loop Projects: Claude Code Prompt prompt used for?
Agent loop memory prompt for writing the AGENTS.md the loop reads every run — context, conventions, architecture, and escalation rules in under 400 words.
How do I get the best results from this memory & state prompt?
Paste the prompt into Claude Code, Cursor, GitHub Copilot, fill in every bracketed placeholder with your specific information, then submit. The more detail you add to each placeholder, the more targeted the output. You can continue in the same conversation to refine tone, length, or focus without restarting from scratch.
Does this prompt work with Claude Code?
Yes. This prompt is optimized for Claude Code and also works well with Cursor and GitHub Copilot. The structured format with a defined role and numbered steps produces consistent results across all major instruction-following models, so you can use whichever tool you prefer.
How do I access this agent loops prompt?
This memory & state prompt is available through TopFreePrompts with a Lucy+ licence. You can get access with a one-time Lucy+ purchase for $109, or a $15/month plan you can cancel anytime. Both options give access to the full library of 1,000,000+ prompts across all categories.
Parallel Worktree Agent Workflow: Claude Code Prompt
Quality Gate Between Maker and Checker: Claude Code Prompt
Verification Agent After Every Pass: Claude Code Prompt
Loop State File Design: Claude Code, Replit Prompt
Three-Agent Pipeline: Explorer, Implementer, Verifier Prompt
Maker-Checker Sub-agent Pair: Claude Code, Cursor Prompt
Goal-Based Loop Until TypeScript Strict: Claude Code Prompt
Nightly Dependency Audit Loop: Claude Code, Replit Prompt
Replace Manual Workflow with Agent Loop: Claude Code Prompt
Minimum Viable Agent Loop: Claude Code, Cursor Prompt
Daily Triage Loop Architecture: Claude Code, Cursor Prompt
Dead Code Elimination Loop: Claude Code, Cursor Prompt
MCP Request and Response Logger and Inspector: Claude Code Prompt
MCP Tool Unit Test Framework: Claude Code, Cursor Prompt
MCP Mock Server for Local Testing: Claude Code, Cursor Prompt
Multi-Tenant MCP Server Design: Claude Code Prompt
MCP Error Handler with Retry Pattern: Claude Code Prompt
MCP Server Health Check Tool: Claude Code, Cursor Prompt
MCP Tool Response Caching Layer: Claude Code Prompt
MCP Rate Limiter Middleware: Claude Code, Cursor Prompt
MCP Server Authentication and Permission System: Claude Code Prompt
Resend Transactional Email MCP Tool: Claude Code Prompt
SMTP Email Sender MCP Tool: Claude Code, Cursor Prompt
Discord Message MCP Tool: Claude Code, Cursor Prompt