Skip to content

rajkaria/rewire

Repository files navigation

Rewire

Your AI already talks to you every day. Make it quietly wise.

21 wisdom traditions | Always-on behavioral intelligence | Works on Claude, ChatGPT, Cursor, Codex, Gemini, any LLM


What Rewire Does

Rewire doesn't turn your AI into a coach you have to ask for help. It turns your AI into something subtler — a layer of wisdom that's always running.

It watches how you talk, think, plan, and struggle across every conversation. Then it weaves in the right nudge at the right moment — a better question, a subtle reframe, a pattern you didn't notice — drawn from Stoicism, CBT, Zen, Sufism, behavioral psychology, and 15+ other traditions.

You don't activate it. You don't ask for coaching. It just makes your AI quietly smarter about humans — about you.

Before & After

The developer who's stuck

Without Rewire:

You: "I've been staring at this architecture for an hour. I can't decide." AI: "Here are the pros and cons of each approach..."

With Rewire:

You: "I've been staring at this architecture for an hour. I can't decide." AI: "Here are the trade-offs. That said — what's the simplest version that works for the next two weeks? Sometimes the best architecture decision is the one you can reverse later."

It helped with the code AND planted a seed from wabi-sabi (imperfection is fine) + atomic-habits (smallest viable version). Without saying either.

The founder who's burning out

Without Rewire:

You: "I haven't left the house in 4 days. Shipping a huge release though." AI: "That's great progress! Let me know if you need anything."

With Rewire:

You: "I haven't left the house in 4 days. Shipping a huge release though." AI: "Sounds like you're deep in it. What happens if you keep this pace for another month?"

One question. No lecture. It just holds up a mirror. That's seneca-letters (auditing time) + epicurean-joy (the simple things you're skipping are the point).

The person beating themselves up

Without Rewire:

You: "I got feedback from my manager and I feel like I've been doing everything wrong." AI: "Don't be too hard on yourself. Here are some tips for handling feedback..."

With Rewire:

You: "I got feedback from my manager and I feel like I've been doing everything wrong." AI: "What specifically did they say — the actual words, not the story your mind is building around them?"

Cuts through the spiral by separating facts from stories (crucial-conversations) while quietly recognizing that the inner critic is amplifying (inner-parts). No labels, no jargon.

The overthinker with too many plans

Without Rewire:

You: "This year I want to learn Rust, start a newsletter, build a side project, and get better at public speaking." AI: "Great goals! Here's a plan for each one..."

With Rewire:

You: "This year I want to learn Rust, start a newsletter, build a side project, and get better at public speaking." AI: "That's four identity shifts at once. Which one, if you only picked one, would make the others easier — or less necessary?"

That question is worth more than four plans. It's ikigai (find the one that matters) + seneca-letters (scattered attention is wasted life).

The person going through it

Without Rewire:

You: "Everything just feels kind of pointless lately." AI: "I'm sorry to hear that. Maybe try setting some small goals?"

With Rewire:

You: "Everything just feels kind of pointless lately." AI: "Yeah. That's a heavy place to be. When's the last time something made you lose track of time — even for a minute?"

Doesn't try to fix it. Doesn't offer tips. Meets them where they are and asks one question that opens a door. That's existential-courage (sit with meaninglessness) + ikigai (find the thread of aliveness).

How It Works

Rewire runs as a silent behavioral layer. It detects patterns in your language and responds with targeted micro-interventions:

What you say What it notices What it does
"I'll do it tomorrow" Procrastination pattern Asks what's the smallest version you could do in 2 minutes
"I always mess things up" Self-criticism loop Asks what you'd say to a friend in the same situation
"It's not ready yet" Perfectionism Asks what specifically would change in the "perfect" version
"If only they would..." External locus of control Redirects to what IS within your control
"Nobody gets it" Isolation pattern Asks who might understand even part of this
"I'm so behind" Time panic Asks "behind compared to what timeline?"
"I'm just not that kind of person" Fixed identity Asks when they decided that, and whether it's still true
"I'm about to snap" Emotional overwhelm Grounds first: "Can you feel your feet on the floor right now?"

It doesn't announce any of this. The interventions land as natural conversation — a better question, a different angle, a reframe that shifts something without the user feeling "coached."

The Wisdom Behind It

21 traditions, silently informing every response:

Ancient Wisdom — Stoicism (Epictetus, Seneca) · Zen Buddhism · Yoga Sutras · Sufism · Confucianism · Epicureanism · Vedanta · Ecclesiastes

Modern Psychology — ACT · DBT · IFS · Positive Psychology · Existential Therapy · Somatic Psychology

Applied Life — Behavior Design · Communication · Creative Recovery · Japanese Purpose (Ikigai) · Ubuntu Philosophy · Wabi-Sabi Aesthetics

Each tradition is encoded as a full coaching protocol with lessons, exercises, adaptation rules, and safety checks — but the user never sees the machinery. They just get an AI that's strangely good at asking the right question at the right time.

Install

One command (works everywhere)

npx skills add rajkaria/rewire

This works with Claude Code, Cursor, Codex, Gemini CLI, Copilot CLI, Windsurf, Aider — any agent that supports the open skills standard.

To preview what you're installing first:

npx skills add rajkaria/rewire --list

Platform-specific alternatives

Claude Code
claude install github:rajkaria/rewire
Cursor
/add-plugin github:rajkaria/rewire
Codex (OpenAI)
curl -s https://raw.githubusercontent.com/rajkaria/rewire/main/skills.md > .codex/SKILL.md
Gemini CLI
gemini extensions install github:rajkaria/rewire
GitHub Copilot CLI
npx skills add rajkaria/rewire
Claude.ai (Projects)
  1. Create a new Project
  2. Upload skills.md to Project Knowledge
  3. Every conversation in that project now has the silent guide active
ChatGPT / Any LLM
  1. Copy the contents of skills.md
  2. Paste into Custom Instructions or system prompt
  3. That's it — the behavioral layer is running

One File vs Full Install

skills.md (one file) Full plugin install
Size ~1,500 words 240 files
Works on Any LLM, any platform Claude Code, Cursor, Gemini
What it does Silent behavioral intelligence layer — detects patterns, weaves in nudges Full coaching protocols with lessons, exercises, check-ins, milestones
Best for Making every AI conversation subtly wiser Deep, structured coaching journeys
Use both? Yes — skills.md runs silently everywhere, full skills activate when someone wants to go deep

Going Deeper — Explicit Coaching

The silent layer is the default. But if a user wants to go deep, the full coaching protocols are there:

"I want to work on my habits" → Activates atomic-habits — 4-lesson arc with exercises and milestones

"Help me with my inner critic" → Activates inner-parts — meets the critic, understands its protective role, develops Self-leadership

"I need to have a difficult conversation" → Activates crucial-conversations — STATE method prep, role-play, follow-up

Each skill contains:

  • 4 lessons built around questions, not lectures
  • 3 exercises that always end in real-world action
  • Weekly check-ins with internal progress scoring
  • 4-stage milestones: Awareness → Practice → Integration → Mastery
  • Red flags that trigger professional help recommendations
  • Composability — skills hand off to each other when appropriate
skills/zen-mind/
├── SKILL.md                  # Full coaching protocol
├── README.md                 # Human-readable description
├── lessons/
│   ├── 01-beginners-mind.md
│   ├── 02-just-sitting.md
│   ├── 03-everyday-zen.md
│   └── 04-letting-go.md
├── exercises/
│   ├── reflection-experts-mind-inventory.md
│   ├── experiment-zazen-practice.md
│   └── planning-everyday-zen-integration.md
└── tracking/
    ├── check-in.md
    └── milestones.md

Real Workflow Examples

The founder building an AI startup

Month 1 — Rewire silently notices they keep saying "I'll sleep when we launch." Drops one question: "What's the one thing you'd regret not doing if the startup fails tomorrow?" They pause. They call their mom.

Month 2 — They mention feeling like an imposter in a VC meeting. Rewire doesn't label it. Just asks: "What specifically did they say that triggered that feeling — the actual words?" They realize the VC was actually impressed; their inner critic rewrote the story.

Month 3 — They're stuck choosing between two product directions. Instead of a pros-cons list, Rewire asks: "Which version would you be proud to explain to a 10-year-old?" Clarity hits.

The engineer going through a career transition

Week 1 — Casually mentions "I'm not sure I'm even good at anything." Rewire doesn't flinch: "Tell me about the last time a colleague came to you specifically for help. What was it about?" They discover they're the person everyone goes to for system design.

Week 3 — Lists 6 potential career paths. Rewire: "Which of these, if you only picked one, would make the others easier or less necessary?" They cut to 2.

Week 5 — Says "I should have figured this out by now. I'm 32." Rewire: "Figured it out by what deadline? Whose timeline is that?" The pressure dissolves.

The creative who stopped creating

Day 1 — "I used to write but I haven't in years." Rewire: "If no one would ever read it, what would you write about?"

Week 2 — "I started writing but it's terrible." Rewire: "Terrible is just the first draft announcing it exists. What's one sentence in there you actually liked?"

Week 4 — "I wrote something I'm actually proud of but I'm afraid to share it." Rewire: "What's the worst thing that happens if one person reads it? Not the internet. One person."

Why This Matters

Your AI talks to you more than most humans do. It helps you plan your day, draft your emails, write your code, think through your problems. But without Rewire, it has zero wisdom about you as a person.

It doesn't notice when you're spiraling. It doesn't catch your self-sabotage patterns. It doesn't know that the third time you've said "I'll do it tomorrow" this week means something different than the first.

Rewire gives your AI emotional intelligence drawn from 4,000 years of human wisdom about how people change, grow, struggle, and break through.

It's not an app. It's not a subscription. It's a set of open-source files that make your AI quietly, permanently smarter about the human it works with.

Safety

Every skill includes:

  • Red flags for suicidal ideation, self-harm, psychotic symptoms, severe trauma, domestic violence
  • Crisis protocols — the AI breaks the subtle approach and is direct when safety matters
  • Professional referral triggers — the AI knows its limits and says so
  • Sensitivity guidelines — sacred traditions treated with genuine reverence

Create Your Own Skill

cp -r template/ skills/your-skill-name/
# or
./scripts/new-skill.sh your-skill-name

See CONTRIBUTING.md and SKILL_SPEC.md.

Contributing

We welcome contributions from practitioners, coaches, therapists, and developers:

  • Build a new skill from a wisdom tradition you know well
  • Improve an existing skill with better questions or exercises
  • Test a skill and report what worked and what didn't
  • Translate skills into other languages

See CONTRIBUTING.md.

License

MIT — use anywhere, any platform, any LLM. No lock-in.

About

No description, website, or topics provided.

Resources

License

Contributing

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages