dennisthiessen/claude-inner-dialog
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add missing reference files and streamline documentation

Browse Source 
- Add missing modality files: lifespan-integration, somatic-experiencing, psychodynamic
- Add missing persona file: grounded-real
- Refactor docs to reduce overlap and clarify focus:
  - README: overview and why (removed install steps, links to other docs)
  - GETTING-STARTED: step-by-step setup only
  - SECURITY: privacy options with friendlier language
- Simplify language for non-technical audience
- Remove duplicate FAQ, crisis resources, and privacy explanations

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Anthony Taglianetti
2026-02-01 21:27:44 -08:00
parent 301dfd3c1e
commit 783fd4a5ae
7 changed files with 538 additions and 443 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+75 -165
View File
@@ -1,8 +1,8 @@
# AI Therapy Starter Kit
A local-first, privacy-focused toolkit for AI-assisted therapy and self-reflection.
A private, persistent AI therapist that lives on your computer.
Your sessions stay on your computer. Your data is yours. Works with Claude, GPT, or any LLM.
Your sessions stay local. Your AI remembers everything. Works with Claude, GPT, or any AI.
---
@@ -10,10 +10,10 @@ Your sessions stay on your computer. Your data is yours. Works with Claude, GPT,
If you've used ChatGPT or Claude for emotional support, you've probably hit these problems:
- **Sessions disappear.** Start a new chat, lose all context. Copy-paste to Google Docs? That gets old fast.
- **Privacy concerns.** Your deepest thoughts, stored on someone else's servers, potentially reviewed by humans.
- **Platform lock-in.** Build a relationship with 4o, then OpenAI deprecates it. Start over.
- **Inconsistent quality.** Sometimes helpful, sometimes reads from a script. No therapeutic framework.
- **Sessions disappear.** Start a new chat, lose all context.
- **Privacy concerns.** Your deepest thoughts on someone else's servers.
- **Platform lock-in.** Build a relationship with one AI, then it changes or goes away.
- **Inconsistent quality.** Sometimes helpful, sometimes generic.
This toolkit solves all of them.
@@ -21,58 +21,25 @@ This toolkit solves all of them.
## What You Get
**Persistent memory** — Sessions and insights accumulate in local markdown files. Your AI therapist remembers everything, across sessions, forever.
**Your AI remembers** — Sessions and insights accumulate in files on your computer. Your AI therapist remembers everything, forever.
**Real privacy** — Files live on your machine. Optional encryption (AES-256) for sensitive data. Use a local LLM for maximum privacy.
**Real privacy** — All files stay on your machine. Optional password protection for shared computers.
**Portable** — Works with Claude Code, ChatGPT API, or any LLM that can read files. Switch providers without losing your history.
**Portable** — Works with Claude, ChatGPT, or any AI that can read files. Switch anytime without losing your history.
**Evidence-based framework** — Built-in CBT, ACT, DBT, Somatic Experiencing, and Lifespan Integration. Structured therapeutic approaches, not just vibes.
**Evidence-based** — Built-in therapeutic approaches (CBT, ACT, DBT, and more). Real frameworks, not just conversation.
**Customizable** — Choose your therapist's communication style. Add modalities. Make it yours.
**Customizable** — Choose your therapist's name and communication style. Make it yours.
---
## Quick Start
## Get Started
**New to Claude Code or terminal-based AI tools?** See the [Getting Started Guide](docs/GETTING-STARTED.md) for step-by-step instructions.
**Ready to set up your AI therapist?** See the [Getting Started Guide](docs/GETTING-STARTED.md) for step-by-step instructions.
### Prerequisites
- [Claude Code](https://claude.ai/code) (recommended) or another LLM interface
- macOS or Windows
- 10 minutes for setup
### Installation
**macOS:**
```bash
git clone https://github.com/ataglianetti/ai-therapy-kit.git
cd ai-therapy-kit
./setup.sh
```
**Windows (PowerShell):**
```powershell
git clone https://github.com/ataglianetti/ai-therapy-kit.git
cd ai-therapy-kit
.\setup.ps1
```
The setup script will guide you through:
1. Choosing a therapist name and communication style
2. Selecting therapeutic approaches (CBT, ACT, DBT skills)
3. Setting up storage location
4. Optional encryption
### Start a Session
```bash
cd ~/ai-therapy # or your chosen location
claude # or your preferred LLM interface
```
Just talk. Your AI therapist will maintain context, take notes, and remember everything for next time.
Setup takes about 10 minutes and requires:
- A computer (Mac or Windows)
- Claude Code (included with Claude Pro, or pay-per-use)
---
@@ -80,178 +47,121 @@ Just talk. Your AI therapist will maintain context, take notes, and remember eve
### Communication Styles
| Style | Best For |
|-------|----------|
| **Warm & Supportive** | Validation-first, gentle challenges, nurturing presence |
| **Direct & Challenging** | Socratic questioning, will push back, insight-focused |
During setup, you choose how your AI therapist communicates:
| Style | Description |
|-------|-------------|
| **Warm & Supportive** | Validation first, gentle challenges, nurturing |
| **Direct & Challenging** | Will push back, asks probing questions |
| **Coach** | Action-oriented, goal-focused, accountability |
| **Grounded & Real** | Down-to-earth, honest feedback, uses humor, focused on growth |
| **Grounded & Real** | Down-to-earth, honest, uses humor |
### Therapeutic Approaches
- **CBT** (Cognitive Behavioral Therapy) — Identify and challenge unhelpful thought patterns
- **ACT** (Acceptance & Commitment Therapy) — Values-based action, mindful acceptance
- **DBT Skills** — Emotional regulation, distress tolerance, interpersonal effectiveness
- **Lifespan Integration** — Body-based trauma integration, builds coherent life narrative
- **Somatic Experiencing** — Nervous system regulation, completing stuck survival responses
Your AI therapist can draw from multiple evidence-based approaches:
### Session Continuity
- **CBT** — Identify and challenge unhelpful thought patterns
- **ACT** — Values-based action, mindful acceptance
- **DBT Skills** — Emotional regulation, distress tolerance
- **Lifespan Integration** — Body-based trauma integration
- **Somatic Experiencing** — Nervous system awareness and regulation
- **Psychodynamic** — Understanding patterns from the past
Your AI therapist automatically:
- Reads your profile and recent sessions at the start
- Takes structured notes at the end
- Updates your profile as insights emerge
- Tracks homework and follows up
### How Sessions Work
### Optional Encryption
Your AI therapist:
- Remembers everything from previous sessions
- Takes notes at the end of each conversation
- Updates your profile as new insights emerge
- Follows up on anything you committed to try
For shared computers or maximum privacy:
- **macOS:** Encrypted disk image (built-in, no install)
- **Windows:** VeraCrypt container (free, open-source)
### Your Files
Setup walks you through it.
---
## How It Works
Everything is stored as simple text files you can read anytime:
```
ai-therapy/
├── CLAUDE.md # Your therapist's personality and approach
├── profile.md # Cumulative client profile (builds over time)
your-therapist-folder/
├── CLAUDE.md # Your therapist's personality
├── profile.md # Your ongoing profile
└── sessions/
├── 2024-01-15.md
├── 2024-01-18.md
└── ...
```
**CLAUDE.md** — Instructions that shape your AI therapist's behavior. Generated during setup based on your preferences.
**profile.md** — Your ongoing profile. Background, patterns, insights, goals. Updated across sessions.
**sessions/** — Session notes in dated markdown files. Themes, emotional state, homework, threads to revisit.
Everything is plain text. Readable, portable, yours.
No special software needed to view your own notes.
---
## Safety Features
This toolkit includes built-in safety protocols:
### Crisis Response
- Recognizes crisis language (suicidal ideation, self-harm, psychosis)
- Provides immediate crisis resources (988, Crisis Text Line)
- Does not attempt to "treat" acute crises
### Scope Limitations
- Will not diagnose conditions
- Will not advise on medication
- Recommends professional help when appropriate
### Ethical Guidelines
- Maintains appropriate therapeutic boundaries
- Promotes autonomy, not dependency
- Acknowledges AI limitations honestly
---
## Important Disclaimer
## Safety & Limitations
**This tool is for self-reflection and emotional support. It is not a replacement for professional mental health care.**
- The AI cannot diagnose mental health conditions
- The AI cannot prescribe or advise on medication
- If you're in crisis, contact a crisis line or emergency services
- Consider sharing that you use this tool with your therapist
Your AI therapist:
- Cannot diagnose mental health conditions
- Cannot prescribe or advise on medication
- Will connect you to crisis resources if needed
- Will recommend professional help when appropriate
**Crisis Resources:**
**If you're in crisis:**
- **988** — Suicide & Crisis Lifeline (call or text, US)
- **741741** — Crisis Text Line (text HOME)
- **911** — Immediate emergencies
- **[findahelpline.com](https://findahelpline.com)** — International directory
- **[findahelpline.com](https://findahelpline.com)** — International
---
## Privacy & Security
## Privacy
### Your Data
All your files stay on your computer. Nothing is uploaded by this toolkit.
- All files stored locally on your machine
- Nothing uploaded, synced, or transmitted by this toolkit
- No analytics, telemetry, or tracking
- Delete anytime by removing the folder
When you chat, your messages go through the AI provider you're using (Claude or ChatGPT). Their privacy policies apply, but API access is more private than their consumer chat apps—your conversations aren't used to train their models.
### API Considerations
If using a cloud LLM (Claude, GPT), your messages go through their API:
- **Claude API:** [Anthropic's privacy policy](https://www.anthropic.com/privacy) — Data not used for training
- **OpenAI API:** [OpenAI's privacy policy](https://openai.com/policies/privacy-policy)
For maximum privacy, use a local LLM (Ollama, LM Studio). See [SECURITY.md](SECURITY.md) for setup.
### Encryption
Optional but recommended for shared computers. Setup script walks you through it.
**Want more privacy options?** See [SECURITY.md](SECURITY.md) for:
- Password-protected folders
- Running a fully local AI (nothing leaves your computer)
---
## Known Limitations
### LLM Safety Guardrails
### AI Safety Filters
All major LLMs (Claude, GPT, etc.) have built-in safety filters. These can affect therapy conversations in ways you should know about:
All major AIs (Claude, ChatGPT) have built-in safety filters that can affect therapy conversations:
**Topics where the AI may seem hesitant or awkward:**
**Topics where the AI may seem hesitant:**
- Sexual trauma and assault
- Abuse (especially detailed descriptions)
- Self-harm (specifics, not general discussion)
- Self-harm specifics
- Violent experiences
- Suicidal ideation (the AI will pivot to crisis resources)
**What this looks like:**
- The AI may give generic responses instead of engaging deeply
- It may redirect the conversation or change the subject
- It may add disclaimers or suggest professional help more than necessary
- In some cases, it may refuse to continue a line of discussion
- Generic responses instead of engaging deeply
- Redirecting the conversation
- Extra disclaimers or suggestions to seek professional help
**This is not the AI judging you.** It's hitting guardrails built into the model. The AI cannot override these filters.
**This is not the AI judging you.** It's hitting automatic filters it can't control.
**What you can do:**
- Frame topics in terms of feelings and impact rather than graphic details
- If the AI redirects, try rephrasing: "I want to process how this affected me emotionally"
- Understand that some trauma processing may require a human therapist who can hold space for the full story
- Consider a local LLM with fewer guardrails (see SECURITY.md), though this trades off response quality
This limitation is real, and we want you to know about it upfront rather than feel dismissed mid-session.
**What helps:**
- Focus on feelings and impact rather than graphic details
- Rephrase: "I want to process how this affected me emotionally"
- Know that some trauma processing needs a human therapist who can hold space for the full story
---
## FAQ
**Can I use this with ChatGPT instead of Claude?**
**Can I use ChatGPT instead of Claude?**
Yes. Copy the contents of your CLAUDE.md file into ChatGPT's Custom Instructions or create a GPT with it. You won't get automatic session saving, but the therapeutic framework works the same.
Yes. The CLAUDE.md file works as a system prompt for any LLM. For ChatGPT, paste it into Custom Instructions or a GPT Project.
**What if I'm already seeing a therapist?**
This tool works well as a supplement between sessions. Consider mentioning it to your therapist.
**What if I'm already in therapy?**
Great! This tool works well as a supplement. Consider sharing that you use it with your therapist.
**Is this HIPAA compliant?**
This is not a healthcare product and doesn't fall under HIPAA. You're running it on your own computer. We don't store or access any of your data.
**Can I customize the therapeutic approach?**
Yes. Edit CLAUDE.md directly, or purchase expansion packs for additional modalities.
**Can I customize my AI therapist?**
Yes. You can edit the CLAUDE.md file directly to adjust the approach, style, or focus areas.
**What if I need more than self-reflection?**
This tool is designed for everyday emotional support. If you're dealing with severe symptoms, trauma, or crisis situations, please seek professional help. The AI will also recommend this when appropriate.
**Why does the AI seem evasive when I talk about trauma?**
LLMs have safety guardrails that can make them awkward around certain topics (abuse, sexual trauma, self-harm, violence). This isn't the AI judging you — it's hitting built-in filters it can't override. See "Known Limitations" above for more detail and workarounds.
This is designed for everyday emotional support. For severe symptoms, trauma, or crisis situations, please work with a professional. Your AI therapist will also suggest this when appropriate.
---