Shadow — Reactive Dog Training Coach

Shadow is an AI-powered reactive dog training coach built on force-free, evidence-based methods. It provides personalised guidance grounded in a curated knowledge base of 26 training topics, 20 breed-specific reactivity profiles, and a three-type reactivity framework (fear / frustration / excitement).

Shadow was created for Calming Paws, a reactive dog training app, and is now available as a standalone skill for any LLM.

Who is this for?

• Dog owners living with a reactive dog who want evidence-based, force-free guidance
• Dog trainers looking for a structured AI assistant to complement their practice
• Developers building pet-care or training apps who want a reactive-dog knowledge layer
• AI enthusiasts who want a well-structured domain-specific skill to study or extend

What Shadow knows

Area	Coverage
Training techniques	DS/CC, LAT, Engage-Disengage, BAT 2.0, pattern games, stationing, muzzle conditioning, emergency protocols
Reactivity framework	Fear-based, frustration/barrier, excitement/over-arousal — identification and treatment
Breed profiles	20 breeds with reactivity types, training nuances, threshold notes, and management tips
Safety	Risk-level escalation, aversive method guardrails, medical redirect, certified trainer referrals
Resources	7 recommended books, 4 trainer credential types, finding a qualified trainer

Install

Shadow is a single Markdown file (SKILL.md). Pick your LLM below and paste it in.

Claude (claude.ai)

Create a Project → paste contents into Custom Instructions.

1. Copy the full contents of SKILL.md
2. Go to claude.ai → Projects → Create project
3. Name it "Shadow" (or whatever you'd like)
4. Open Project knowledge / Custom Instructions and paste the contents
5. Start a new chat inside the Project — Shadow is now your reactive-dog coach

ChatGPT (chat.openai.com) - https://chatgpt.com/g/g-6a10e840d2708191b532fabacd01e27f-shadow

Cursor / Continue (IDE)

Add to .cursorrules (Cursor) or .continuerules / config (Continue).

1. Clone or download this repo
2. Copy SKILL.md into your project root as .cursorrules (Cursor) or append it to your Continue config's systemMessage

   # Cursor
   cp shadow-reactivity-coach/SKILL.md /path/to/your/project/.cursorrules
   
   # Continue — add to ~/.continue/config.json under "systemMessage"

3. Restart the IDE so the new rules load
4. Shadow's guidance now travels with that workspace

Claude Code (Skills directory)

git clone https://github.com/codeWithRewaskar/shadow-reactivity-coach.git
mkdir -p .claude/skills
cp shadow-reactivity-coach/SKILL.md .claude/skills/shadow-reactive-coach.md

Examples

The examples/ directory contains 10 real conversations showing Shadow in action:

#	Scenario	Key concepts
1	What is reactivity?	Three types, not aggression, threshold zones
2	My dog lunges at other dogs	Frustration vs fear diagnosis, Engage-Disengage
3	Border Collie threshold work	Breed-specific advice, arousal management
4	Emergency reaction on a walk	U-turn, scatter feeding, cortisol recovery
5	Is it fear or frustration?	Diagnostic framework, body language
6	Training has stalled	Cortisol vacation, plateau management
7	Just adopted a rescue	3-3-3 rule, decompression, trigger identification
8	Starting muzzle training	Conditioning steps, basket muzzle, timeline
9	Small dog fear reactivity	Chihuahua-specific, choice-based greetings
10	Setback after progress	Non-linear progress, trigger stacking, recovery

Project structure

shadow-reactivity-coach/
├── SKILL.md          # The complete skill — paste into any LLM
├── README.md         # You are here
├── LICENSE           # CC-BY-4.0
├── CHANGELOG.md      # Version history (currently v2.1)
└── examples/         # 10 conversation examples

Principles

• Force-free only. Shadow never recommends aversive tools or punishment-based methods. Hard guardrails block these at the prompt level.
• Evidence-based. All knowledge is sourced from vetted force-free training resources and certified bodies (IAABC, CCPDT, KPA).
• Breed-aware. 20 breed profiles ensure advice is tailored, not generic.
• Safe escalation. High-risk situations always include a certified trainer referral.
• Type-aware. The three-type reactivity framework (fear / frustration / excitement) runs through every recommendation.

Contributing

Contributions welcome! If you'd like to:

• Add a breed profile — follow the schema in SKILL.md and include: reactivity type, common triggers, training nuances, management tips, and threshold notes.
• Add a training topic — source must be from a certified force-free body or evidence-based reference. No aversive methods, no dominance framing.
• Fix an error — open an issue or PR with the correction and your source.

License

CC-BY-4.0 — free to use, share, and adapt with attribution.

Built by Saurabh Rewaskar as part of Calming Paws.

---

🐾 Powered by Calming Paws

Shadow is the open-source coaching brain behind Calming Paws — the full reactive-dog companion app with live dog profiles, walk logs, trigger tracking, and progress analytics built right in. If you'd like the complete experience (not just the LLM prompt), head to calming-paws.com.