Structured Coding with AI — Open Paws

Ready-to-use instruction file sets for 12 AI coding tools, built for developers working on animal advocacy software. Copy one directory into your project root and your AI assistant immediately understands the advocacy domain: three-adversary security threat model, activist identity protection, farmed animal terminology, and workflow standards. No external dependencies — just files.

Maintained by Open Paws, a 501(c)(3) nonprofit building AI infrastructure for the animal liberation movement.

Note

This project is part of the Open Paws ecosystem — AI infrastructure for the animal liberation movement. Explore the full platform →

Directory Structure

structured-coding-with-ai/
├── agents-md/          # AGENTS.md — vendor-neutral, 20+ tool support
├── aider/              # CONVENTIONS.md
├── augment-code/       # .augment/rules/ (14 files)
├── claude-code/        # CLAUDE.md + .claude/rules/ + .claude/skills/ (17 files)
├── cline/              # .clinerules/ (14 files)
├── cursor/             # .cursorrules + .cursor/rules/ (14 files)
├── gemini-cli/         # GEMINI.md
├── github-copilot/     # .github/ instructions, prompts, chat modes (23 files)
├── jetbrains-junie/    # .junie/guidelines.md
├── kilo-code/          # .kilocode/rules/ + Memory Bank (21 files)
├── roo-code/           # .roomodes + .roo/rules/ (19 files)
├── windsurf/           # .windsurf/rules/ (14 files)
└── scripts/            # Unicode integrity checker

Quickstart

Clone the repo

git clone https://github.com/Open-Paws/structured-coding-with-ai.git

Pick your tool from the table below and copy its directory into your project root

Claude Code example

cp claude-code/CLAUDE.md your-project/
cp claude-code/hooks-template.md your-project/
cp -r claude-code/.claude your-project/

Add your project specifics — build commands, language conventions, tech stack. The instruction files provide methodology; they do not replace project-specific setup.
Verify integrity before committing instruction file edits
```
python scripts/check-unicode-integrity.py
```

Full copy commands for every tool are listed in the Tool Reference section below.

Tools Covered

Tool	Directory	Files	Format
Claude Code	`claude-code/`	17	CLAUDE.md + scoped rules + skills + hooks template
Cursor	`cursor/`	14	.cursorrules + .mdc files with 4 activation modes
GitHub Copilot	`github-copilot/`	23	copilot-instructions.md + instructions + prompts + chat modes + skills
Windsurf	`windsurf/`	14	.windsurf/rules/ with 4 trigger types, within 6K/12K char limits
Kilo Code	`kilo-code/`	21	Mode files + Memory Bank + concern files + skills
Cline	`cline/`	14	.clinerules/ with Plan/Act paradigm
Roo Code	`roo-code/`	19	.roomodes JSON + mode rules + concern files + skills
Augment Code	`augment-code/`	14	.augment/rules/ directory
Aider	`aider/`	1	Single CONVENTIONS.md
Gemini CLI	`gemini-cli/`	1	Single GEMINI.md
JetBrains / Junie	`jetbrains-junie/`	1	Single .junie/guidelines.md, always loaded
AGENTS.md	`agents-md/`	1	Vendor-neutral, supported by 20+ tools
Total		140

Content Coverage

Every tool directory covers the same material, adapted to its native format. Tools with multi-file support split content into separate files with appropriate activation triggers. Single-file tools condense everything into clearly-headed sections.

7 Concerns (in every tool)

Testing — Spec-first generation, mutation testing, assertion quality over coverage percentage
Security — Three-adversary threat model (state surveillance, industry infiltration, AI model bias), zero-retention APIs, supply chain verification, device seizure preparation
Privacy — Activist identity protection, coalition data sharing, whistleblower protection, real deletion (not soft delete)
Cost optimization — Model routing, token budgets, prompt caching, vendor lock-in as movement risk
Advocacy domain — Ubiquitous language dictionary, bounded contexts for investigation ops / public campaigns / coalition coordination / legal defense
Accessibility — i18n from day one, offline-first, low-bandwidth, low-literacy design
Emotional safety — Progressive disclosure of traumatic content, configurable detail levels, secondary trauma mitigation for developers

6 Process Skills (invoked on demand)

git-workflow — Issue-first GitHub workflow with worktree-per-task, plan-review-implement loop, desloppify gate
testing-strategy — Spec-first generation, five anti-patterns to avoid, mutation-guided improvement
requirements-interview — Structured stakeholder questions covering threat model, coalition needs, user safety
plan-first-development — Spec, design, decompose, implement one subtask at a time
code-review — Layered review pipeline, Ousterhout red flags, AI-specific failure patterns, advocacy data leak checks
security-audit — Advocacy threat model assessment, slopsquatting defense, prompt injection detection, MCP server security

External Contribution Safety

All 12 directories include a two-state identity model: advocacy mode for Open Paws repos (full domain context active) and neutral mode for external open-source contributions (org identity suppressed, no advocacy framing in commits or PRs). This prevents AI agents from inadvertently identifying contributors or leaking organizational context when contributing to third-party projects.

Tool Reference

Copy commands for all 12 tools

# Claude Code
cp claude-code/CLAUDE.md your-project/
cp claude-code/hooks-template.md your-project/
cp -r claude-code/.claude your-project/

# Cursor
cp cursor/.cursorrules your-project/
cp -r cursor/.cursor your-project/

# GitHub Copilot
cp -r github-copilot/.github your-project/

# Windsurf
cp -r windsurf/.windsurf your-project/

# Kilo Code
cp -r kilo-code/.kilocode your-project/

# Cline
cp -r cline/.clinerules your-project/

# Roo Code
cp roo-code/.roomodes your-project/
cp -r roo-code/.roo your-project/

# Augment Code
cp -r augment-code/.augment your-project/

# Aider
cp aider/CONVENTIONS.md your-project/

# Gemini CLI
cp gemini-cli/GEMINI.md your-project/

# JetBrains / Junie
cp -r jetbrains-junie/.junie your-project/

# AGENTS.md (universal fallback, 20+ tool support)
cp agents-md/AGENTS.md your-project/

Architecture

How instruction files are structured across tools

Each tool directory is self-contained and independently authored for its format — not a template applied 12 times.

Multi-file tools (Claude Code, Cursor, GitHub Copilot, Windsurf, Kilo Code, Cline, Roo Code, Augment Code) break content into separate files with activation triggers:

Always-on files: core methodology, ubiquitous language, security baseline
Conditionally loaded files: concern-specific rules triggered by file path or AI decision
On-demand files: process skill guides invoked explicitly

Single-file tools (Aider, Gemini CLI, JetBrains/Junie, AGENTS.md) condense all 7 concerns and 6 skills into clearly-headed sections within one file.

GitHub Copilot has the richest hierarchy: copilot-instructions.md (repo-wide), instructions/ (path-specific with applyTo:), prompts/ (reusable workflows), chat-modes/ (persistent agent personas), and skills/.

Roo Code uses .roomodes JSON to define custom modes (Review, Interview) with tool restrictions and model assignments per mode.

Windsurf enforces hard limits of 6,000 characters per file and 12,000 characters combined across loaded files. Always-on files are budgeted to ~8K total to leave headroom for conditionally loaded files.

Integrity protection: All instruction files are checked on every PR for hidden Unicode characters (the Rules File Backdoor attack). Run python scripts/check-unicode-integrity.py locally before pushing any edits to instruction files.

Ecosystem position: This repo is the canonical AI methodology source for all Open Paws projects. It feeds two downstream consumers:

graze-cli — bundles selected instruction files into its binary
desloppify — generates agent skill files from the same content

structured-coding-with-ai  (canonical AI methodology)
        |
        +-- graze-cli          bundles instruction files into CLI binary
        |
        +-- desloppify         generates agent skill files from content
        |
        +-- gary               autonomous agent uses these as base instructions
        |
        +-- platform           copies claude-code/ into project root
        |
        +-- (all Open Paws     copy relevant tool directory into
             project repos)    their own project roots

Documentation

Claude Code directory README — file-by-file description and copy commands
scripts/check-unicode-integrity.py — integrity checker for instruction files
Open Paws developer docs — platform documentation

Code Quality

Contributing

The most valuable contribution is a new tool directory. If you use an AI coding tool not listed here, add it:

Create tool-name/ at the repo root
Implement all 7 concerns in the tool's native format
Implement all 6 process skills in the tool's native format
Add external contribution safety content (two-state identity model)
Add a README.md inside the directory with file descriptions and copy commands
Add a row to the tools table in this README
Run python scripts/check-unicode-integrity.py before opening a PR
Each tool must be independently authored for its format — do not copy content verbatim from another tool and relabel it

Each tool must cover advocacy-domain content: farmed animal terminology, the three-adversary security model, activist identity protection, and emotional safety for developers. Generic instruction files without advocacy context do not meet the bar for this repo.

License and Acknowledgments

Content across all 12 tool sets was derived from a knowledge base covering empirical research on AI-assisted development, software design principles (Ousterhout, Feathers, DDD), testing strategy, git workflow for AI-generated code, and operational security for the animal advocacy domain.

tech_stack: [markdown, python]
project_status: stable
difficulty: beginner
skill_tags: [ai-coding, instruction-files, animal-advocacy, security, developer-tooling]
related_repos:
  - https://github.com/Open-Paws/graze-cli
  - https://github.com/Open-Paws/desloppify
  - https://github.com/Open-Paws/gary
  - https://github.com/Open-Paws/platform
  - https://github.com/Open-Paws/no-animal-violence

Donate · Discord · openpaws.ai · Volunteer

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Structured Coding with AI — Open Paws

Directory Structure

Quickstart

Tools Covered

Content Coverage

7 Concerns (in every tool)

6 Process Skills (invoked on demand)

External Contribution Safety

Tool Reference

Architecture

Documentation

Code Quality

Contributing

License and Acknowledgments

FilesExpand file tree

README.md

Latest commit

History

README.md

File metadata and controls

Structured Coding with AI — Open Paws

Directory Structure

Quickstart

Tools Covered

Content Coverage

7 Concerns (in every tool)

6 Process Skills (invoked on demand)

External Contribution Safety

Tool Reference

Architecture

Documentation

Code Quality

Contributing

License and Acknowledgments