maind
Your AI forgets everything when you close the chat. Maind fixes that.
One CLI. Every decision, bug, and solution — stored, encrypted, permanent.
Your AI tools plug in and remember across sessions. You never re-explain context again.

MAIND — krnl
encrypted · unlocked · 42 memories · 18 tags · 7 links
── Activity ──
10:32:15 STORE decision: Use JWT with refresh tokens [a3f8c912]
10:31:02 RECALL "auth strategy" → 3 found
10:28:44 SCAN_DONE entries=12 project=myapp [0f4f4f98]
10:15:33 STORE solution: Fixed OOM with streaming parser [b7e2d1a0]
──────────────────────────────────────────────
maind>
The problem
You fix a critical bug. You choose JWT over sessions. You discover a deployment gotcha.
Next week, new chat — your AI has no idea. You re-explain everything. Again.
Your knowledge is scattered across Slack threads, git commits, stale docs, and closed tabs.
None of your AI tools can access any of it.
What Maind does
| Without Maind |
With Maind |
| AI forgets everything per session |
AI remembers across all sessions |
| Context lost in Slack/docs/git |
One searchable, encrypted brain |
| Re-explain the same decisions |
Decisions stored once, recalled forever |
| "Why did we choose X?" — nobody knows |
maind recall "why X" → instant answer |
| New project, blank slate |
Global memory carries across projects |
Install
# Go (1.21+)
go install github.com/rufus-SD/maind@latest
# or clone and build
git clone https://github.com/rufus-SD/maind.git
cd maind && make install
# Python (downloads binary, no Go needed)
git clone https://github.com/rufus-SD/maind.git
cd maind && pip install -e python/
Try it in 30 seconds
maind demo
maind demo creates a sample brain with real-looking memories and opens the dashboard.
No setup, no passphrase — just see what it does.
Real setup (2 minutes)
# 1. Initialize — pick a name, enable encryption
maind init
# 2. Connect your AI
maind connect cursor # or: claude, windsurf, copilot, aider, generic
# 3. Launch the dashboard
maind
That's it. Your AI now has persistent memory. It stores decisions and solutions silently as you work. Next session, it remembers.
Deep scan — the killer feature
Tell your AI:
"scan this project"
It will analyze your codebase, think out loud, and populate your memory with what matters:
maind> scan show 0f4f
Scan [0f4f4f98]
Project: myapp
Status: COMPLETED
Duration: 3m12s
Entries: 12
Thought log:
[10:25:23] Analyzing go.mod — 14 direct deps, none pinned. Risky for reproducible builds.
[10:26:01] Found 3 SQL migrations with no rollback. Recovery from bad deploys is impossible.
[10:27:15] README mentions Redis but no config in deployment manifests.
Memories created:
[a3f8c912] decision — JWT with refresh tokens for auth
[b7e2d1a0] bug — Migration 002 has no rollback path
[c4d5e6f7] learning — Redis cache is documented but never deployed
The AI does the thinking. Maind stores the results. Every thought is encrypted and auditable.
maind connect cursor # Cursor IDE
maind connect claude # Claude Code
maind connect windsurf # Windsurf
maind connect copilot # GitHub Copilot
maind connect aider # Aider
maind connect generic # Any other tool
One command. The AI reads your memories at session start, stores new ones as it works, and searches when it needs context. No config files, no API keys.
How it works
You (or your AI) ──→ maind remember "..." ──→ encrypted SQLite
↓
You (or your AI) ──→ maind recall "..." ←── search + decrypt
- Single binary — no runtime deps, no Docker, no cloud
- AES-256-GCM encryption with Argon2id key derivation
- FTS5 full-text search across all memories
- Session management — enter passphrase once, work freely, auto-locks on exit
- Wrong passphrase rejection — verification token prevents silent corruption
Commands
| Command |
What it does |
maind |
Launch the live dashboard |
maind demo |
Try it instantly with sample data |
maind init |
Guided setup wizard |
maind remember "..." |
Store a memory |
maind recall "..." |
Search memories |
maind list |
List recent memories |
maind show <id> |
Full detail view |
maind forget <id> |
Archive (soft delete) |
maind scan start/log/complete |
Manage AI scan sessions |
maind scan list / show <id> |
Review scans + AI thought logs |
maind connect <tool> |
Wire into an AI assistant |
maind status |
Check if session is ready |
maind unlock / lock |
Manage encryption session |
maind link <from> <to> |
Link related memories |
maind tags |
List all tags |
maind export |
Export as JSON |
Memory kinds
decision · solution · bug · learning · context · snippet · note
Data storage
Everything local. Single SQLite file. Your machine, your data.
| OS |
Default path |
| macOS |
~/Library/Application Support/maind/ |
| Linux |
~/.local/share/maind/ |
| Windows |
%APPDATA%/maind/ |
Contributing
See CONTRIBUTING.md.
Security
See SECURITY.md for the encryption model and vulnerability reporting.
License
MIT