Prompts

A library of production-tested prompt templates for AI-assisted development. These are not chat prompts — they’re structured instruction sets designed to produce consistent, high-quality outputs from AI assistants.

What’s included

Audit MAP methodology (new)

Field-tested execution patterns from 4 real source code audits. The most operationally mature part of this library.

Component	Purpose
Audit MAP Methodology	Recon-first, multi-pass, Finding Contract, validation rigor, attack chains
JSON Sidecar Pattern	AI produces structured JSON alongside human-readable markdown
Audit MAP templates	Ready-to-use templates: recon, single-pass, multi-pass orchestrator, pass, synthesis

Audit prompts

Multi-lens codebase analysis prompts that delegate to specialist sub-agents:

Prompt	Purpose	Use when
codebase-audit-prompt.md	Full codebase audit with 8 lenses	Starting a new project review
claude-code-audit-prompt.md	Claude Code-specific audit	Auditing Claude Code output
claude-code-audit-with-logging.md	Audit with detailed logging	Need an audit trail
dev_audit_prompt.txt	Dev velocity audit	Quick development health check
full_audit_prompt.txt	Comprehensive deep dive	Detailed analysis

See the Audit Framework for lens architecture details.

Methodology guides

Guide	Purpose
CAP Methodology	Develop prompts in modules, deploy as a monolith
god-prompt-methodology.md	Building effective large-scale prompts (MAP design)
Prompt Testing	Testing prompts in Python, TypeScript, and Rust

Standalone audit prompts

Prompt	Purpose
dev-blockers-audit.md	Identify development blockers and friction points
full-production-audit.md	Production readiness assessment

How to choose

Need	Use
Quick dev health check	dev_audit_prompt.txt
Standard project audit (1-5 lenses)	audit/templates/single-pass-template.md
Complex audit (6+ lenses)	audit/templates/multi-pass-orchestrator-template.md
AI-generated code review	claude-code-audit-prompt.md
Machine-readable output	Add JSON sidecar to any audit
Production readiness	full-production-audit.md

How to use them

These are templates — copy and customise:

git clone https://github.com/ydun-code-library/Ydun_ai_workflow.git
cd Ydun_ai_workflow

# Standard audit setup
cp prompts/audit/templates/recon-template.md ./my-audit/00-recon.md
cp prompts/audit/templates/single-pass-template.md ./my-audit/01-audit.md
# Fill placeholders, run recon first, then audit

# Add JSON output
# Include json-sidecar-pattern.md instructions in your audit MAP

Prompt locations

prompts/
├── audit/
│   ├── codebase-audit-prompt.md
│   ├── claude-code-audit-prompt.md
│   ├── claude-code-audit-with-logging.md
│   ├── dev_audit_prompt.txt
│   ├── full_audit_prompt.txt
│   ├── json-sidecar-pattern.md
│   └── templates/
│       ├── recon-template.md
│       ├── single-pass-template.md
│       ├── multi-pass-orchestrator-template.md
│       ├── pass-template.md
│       └── synthesis-template.md
├── methodology/
│   ├── audit-map-execution-patterns.md
│   ├── cap-workflow-methodology.md
│   ├── god-prompt-methodology.md
│   └── prompt-testing-implementation-guide.md
├── dev-blockers-audit.md
└── full-production-audit.md