AGENTS.md is a project-level configuration file that provides AI coding assistants with complete project context, development principles, and operational guidelines. It follows the agents.md standard and works across Claude, Cursor, GitHub Copilot, and other AI tools.

What is Jimmy's Workflow?

Jimmy's Workflow is a four-phase validation system (PRE-FLIGHT, IMPLEMENT, VALIDATE, CHECKPOINT) designed to prevent AI hallucination and ensure robust implementation. Each phase has specific gates and confidence levels (HIGH/MEDIUM/LOW) that determine when human review is needed.

How do I set up multiple AI instances as a team?

Assign each AI instance a role card with identity, responsibilities, personality traits, and success criteria. Use file-based handoff protocols for coordination. Personality orchestration at the system prompt level produces measurably better results than treating AI instances as generic tools.

Philosophy

Ydun AI Workflow exists because AI assistants are powerful but forgetful. Every new session starts with zero context. These templates solve that by giving AI assistants structured project knowledge from the first prompt.

The core problem

AI coding assistants have three failure modes:

Context loss — Every new session starts blank. Decisions, architecture, and conventions are lost between conversations.
Hallucination without checkpoints — AI will confidently claim something works without verifying. No built-in validation gates.
Inconsistency — Without shared principles, each session invents its own approach. Code style, testing strategy, and documentation quality vary wildly.

How templates solve it

Problem	Solution
Context loss	AGENTS.md — AI reads project context on session start
No validation	Jimmy’s Workflow — Mandatory PRE-FLIGHT → VALIDATE → CHECKPOINT gates
Inconsistency	11 Core Principles — Same rules every session, even after context compression
Session breaks	STATUS.md + NEXT-SESSION-START-HERE.md — Continuity between sessions
Poor documentation	Documentation Standards — AI-optimized, structured, searchable

Design principles

Documentation-first. Write the specification before the implementation. If it’s not documented, it doesn’t exist.

Test-driven. Write the test first. Run it to see it fail. Then write the code. This applies to AI-assisted development as much as traditional development — arguably more, because AI is more likely to claim something works without checking.

Local-first. Your project context stays in your repo. No external services, no API calls, no vendor lock-in. AGENTS.md is a markdown file in your project root.

Pragmatic. Working solutions over elegant theory. These templates evolved from real projects over months of daily use, not from a whiteboard exercise.

Quality paramount. Measure twice, cut once. The templates enforce this through Jimmy’s Workflow confidence levels — HIGH confidence proceeds automatically, MEDIUM pauses for human review, LOW stops for human validation.

What these templates are NOT

Not a framework or library — they’re markdown files
Not tied to any specific AI assistant — they work with Claude, Cursor, Copilot, Gemini, or anything that reads project files
Not auto-generated — you fill in your project’s actual context
Not prescriptive about tech stack — they define workflow and principles, not implementation choices

Evolution

These templates have been in daily production use since late 2024. They’ve gone through five stages:

v0: Copy-pasting ChatGPT output into VS Code
v1: IDE-integrated AI (Cursor)
v2: CLI tools, structured files, validation gates
v3: Multi-machine coordination, role specialisation
v4: Multi-agent team with personalities, handoff protocols, dead drops

See the Evolution guide for the full journey.