https://github.com/kbanc85/claudia
Terminal-based AI chief of staff. Remembers relationships, tracks commitments, helps you think strategically. Runs on Claude Code.
https://github.com/kbanc85/claudia
ai-assistant claude-code productivity relationship-management terminal
Last synced: 3 days ago
JSON representation
Terminal-based AI chief of staff. Remembers relationships, tracks commitments, helps you think strategically. Runs on Claude Code.
- Host: GitHub
- URL: https://github.com/kbanc85/claudia
- Owner: kbanc85
- License: other
- Created: 2026-01-23T22:38:14.000Z (23 days ago)
- Default Branch: main
- Last Pushed: 2026-02-08T10:32:32.000Z (8 days ago)
- Last Synced: 2026-02-08T10:53:29.756Z (8 days ago)
- Topics: ai-assistant, claude-code, productivity, relationship-management, terminal
- Language: Python
- Size: 1.67 MB
- Stars: 52
- Watchers: 1
- Forks: 6
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: CHANGELOG.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
- Code of conduct: CODE_OF_CONDUCT.md
- Security: SECURITY.md
- Notice: NOTICE
Awesome Lists containing this project
README
A thinking partner who tracks relationships, not just tasks.
Catches commitments. Remembers context. Connects the dots across your network.
Try the Demo ·
Why Claudia ·
Install ·
How It Works
---
## The Problem With AI Assistants
You tell ChatGPT about Sarah on Monday. By Wednesday, it's forgotten.
You make a promise in a meeting. Nobody tracks it. You promise a deliverable on Friday and lose track of it by Wednesday. You meet someone important, and three weeks later you can't remember what you talked about.
**AI tools don't have memory. Claudia does.**
---
## What Makes Claudia Different
🎯 Catches Commitments
Say "I'll send that by Friday" and she tracks it. On Friday morning, she reminds you.
🔗 Remembers Relationships
Mention Sarah from Acme and Claudia surfaces: last conversation, open promises, sentiment.
⚠️ Warns Before Things Slip
Haven't talked to your best client in 3 weeks? She tells you before it's a problem.
📄 Shows Her Work
Every fact traces to its source. Ask "how do you know that?" and she shows the receipt.
🧠 Visualize Your Memory
Run /brain to see a 3D graph of everything she knows: people, projects, connections.
🔒 Fully Local
Everything runs on your machine. Your data never leaves. Delete anything, anytime.
---
## Try It in 30 Seconds
Demo mode creates a pre-populated installation with realistic fake data. No setup, no configuration.
```bash
npx get-claudia my-demo --demo
cd my-demo
claude
```
**What's in the demo:**
- 60 people across investor, founder, and client networks
- 15 organizations and 15 projects
- 115 memories (facts, commitments, observations)
- Overdue items and relationship warnings to explore
- Full 3D brain visualization
The demo database is isolated in `~/.claudia/demo/`. Your real data is never touched.
---
## Quick Start
```bash
npx get-claudia
cd claudia
claude
```
Say hi. She'll introduce herself, learn about you in a natural conversation, and generate a personalized workspace within a few sessions.
**Requirements:** [Claude Code](https://docs.anthropic.com/en/docs/claude-code), Node.js 14+, Python 3.10+ (for memory)
Upgrading from a previous version?
```bash
cd your-claudia-directory
npx get-claudia .
```
This upgrades framework files while preserving your data (context/, people/, projects/).
Add Brain Visualizer to existing install
```bash
# macOS/Linux
bash "$(npm root -g)/get-claudia/visualizer/scripts/install.sh"
# Windows (PowerShell)
& "$(npm root -g)\get-claudia\visualizer\scripts\install.ps1"
```
---
## See It in Action
```
You: What do I know about Sarah Chen?
Claudia: Sarah Chen is CEO at Meridian Ventures. You met at SaaStr 2025.
📊 Relationship Health: Strong (last contact 3 days ago)
🎯 Open Commitments:
• Send updated pitch deck to Sarah [OVERDUE - was due Friday]
💡 Recent Context:
• She's interested in AI infrastructure angle
• Wants to see technical architecture doc
• Bringing in technical partner for next meeting
🔗 Connected to: Series A (project), Priya Sharma (introduced by)
Source: Call notes from Jan 28, Email thread Jan 30
```
From here, every conversation builds on the last. She remembers.
---
## Who It's For
Claudia detects your work style and generates structure that fits:
| Role | What You Get |
|------|--------------|
| **Consultant** | Client folders, deliverable tracking, pipeline commands |
| **Executive** | Direct reports, initiatives, board prep, leadership tools |
| **Founder** | Investor CRM, fundraising cycles, team and product tracking |
| **Solo Professional** | Flexible structure for wearing many hats |
| **Creator** | Audience growth, content calendar, collaboration tracking |
---
## Key Commands
| Command | What It Does |
|---------|--------------|
| `/morning-brief` | What needs attention today: commitments, meetings, warnings |
| `/brain` | Launch 3D memory visualization |
| `/meeting-prep [person]` | One-page briefing before a call |
| `/capture-meeting` | Process notes into decisions, commitments, action items |
| `/what-am-i-missing` | Surface risks, overdue items, cooling relationships |
| `/memory-audit` | See everything Claudia knows, with source chains |
All commands
| Command | What It Does |
|---------|--------------|
| `/weekly-review` | Guided reflection across relationships and projects |
| `/accountability-check` | Outstanding commitments and waiting-on items |
| `/draft-reply` | Email response drafts in your voice |
| `/follow-up-draft [person]` | Post-meeting thank-you and summary |
| `/new-person [name]` | Create a relationship file |
| `/pipeline-review` | Active deals, capacity, pipeline health |
| `/client-health` | Status across all client relationships |
---
## How It Works
Claudia has two layers:
**Template layer** (markdown) defines who she is. Skills, commands, rules, and identity files that Claude reads on startup.
**Memory system** (Python) defines what she remembers. SQLite + vector embeddings + three services:
| Service | What It Does |
|---------|--------------|
| **Remember** | Stores facts, entities, relationships with embeddings |
| **Recall** | Retrieves via hybrid ranking (vector + importance + recency) |
| **Consolidate** | Background: decay old memories, detect patterns, generate predictions |
```
You ──► Claude Code ──► Reads Claudia's templates ──► Becomes Claudia
│
▼
Memory daemon (local) ◄── MCP tools
│
▼
SQLite + vectors + Ollama (all local)
```
Technical deep dive
**Semantic search** uses hybrid ranking: 50% vector similarity, 25% importance, 10% recency, 15% full-text. Accessing a memory boosts it (rehearsal effect).
**Document storage** keeps files, transcripts, emails on disk, linked to people and memories. Content-hash deduplication. Automatic lifecycle management.
**Provenance chains** trace any fact to its source email, transcript, or conversation.
**Graph traversal** connects dots across your network. Ask about one person, see related entities with top memories.
**Per-project isolation** keeps work memories separate from personal. Each workspace gets its own database.
**Session narratives** capture tone and emotional context, not just facts.
For full architecture diagrams, see [ARCHITECTURE.md](ARCHITECTURE.md).
Cognitive tools (local LLM extraction)
Paste a meeting transcript. A local language model extracts structured data (entities, facts, commitments) in seconds. Claude reviews and applies judgment.
- Runs locally via [Ollama](https://ollama.com), no API keys
- Models: Qwen3-4B (recommended), SmolLM3-3B, Llama 3.2-3B
- Falls back gracefully when no model installed
Four extraction modes: **meeting**, **email**, **document**, **general**.
---
## Privacy and Safety
- **Fully local.** Memory, embeddings, cognitive tools run on your machine. No external APIs for storage.
- **No external actions without approval.** Every email, calendar event, external action requires your explicit "yes."
- **Your data is yours.** `~/.claudia/memory/` (SQLite), `~/.claudia/files/` (documents), readable markdown. Delete anything.
---
## System Requirements
| Component | Required | Purpose |
|-----------|----------|---------|
| [Claude Code](https://docs.anthropic.com/en/docs/claude-code) | Yes | Runtime |
| Node.js 14+ | Yes | Installer |
| Python 3.10+ | Recommended | Memory system |
| [Ollama](https://ollama.com) | Recommended | Embeddings + cognitive tools |
Without the memory system, Claudia still works using markdown files. With it, she gains semantic search, pattern detection, and predictions.
**Platforms:** macOS, Linux, Windows
---
Troubleshooting
**Memory tools not appearing?**
```bash
~/.claudia/diagnose.sh
# Then restart Claude Code in a NEW terminal
```
**Check daemon health:**
```bash
curl http://localhost:3848/health
tail -f ~/.claudia/daemon-stderr.log
```
**Ollama not running after reboot?**
```bash
launchctl load ~/Library/LaunchAgents/com.ollama.serve.plist
```
**Pull models manually:**
```bash
ollama pull all-minilm:l6-v2 # Embeddings
ollama pull qwen3:4b # Cognitive tools (optional)
```
---
## Contributing
Claudia is source-available under the PolyForm Noncommercial License.
- **Template changes:** `template-v2/`
- **Memory system:** `memory-daemon/` (tests: `pytest tests/`)
- **Architecture:** [ARCHITECTURE.md](ARCHITECTURE.md)
- **Dev guide:** [CLAUDE.md](CLAUDE.md)
---
## License
[PolyForm Noncommercial License 1.0.0](https://polyformproject.org/licenses/noncommercial/1.0.0/)
**Free for:** Personal use, research, education, nonprofits.
**Requires permission:** Commercial use. Contact [mail@kbanc.com](mailto:mail@kbanc.com)
---
Created by Kamil Banc · @kamilbanc · AI Adopters Club
"I learn how you work. Let's get started."
If Claudia helps you, a ⭐ on GitHub means a lot.