rsn-db 0.4.6

Rust-powered embedded database for Python: encryption, GraphRAG, MemPalace AI memory, and an agent-friendly CLI

RSN DB

A Rust-powered embedded database for Python — fast, secure, and optionally opinionated.

Installation · Quick start · Documentation · Security · Changelog

---

RSN DB is a single-file embedded database with a native Rust engine and a Python API. Use it like a lightweight local datastore, a graph-aware knowledge store (GraphRAG), or an agent-friendly shell with optional MemPalace AI memory integration.

Highlights

• Performance — Rust core, zstd compression, optimized indexes
• Security — AES-256-GCM encryption at rest, SHA-256 integrity checks, path guards, DoS limits
• Persistence — Explicit save() / load() / snapshot() with JSON engine snapshots
• GraphRAG — Ingest text and query relationships without an external LLM
• Personality modes — Professional, Friendly, or Snarky CLI feedback (130+ snark lines, mood/vitals in Snarky mode)
• AI memory — Session memory sidecar plus optional official MemPalace bridge
• Agent-friendly CLI — Non-interactive -c, --json, --no-prompt for automation

---

Installation

pip install rsn_db

Optional extras

# Official MemPalace integration (local-first AI memory)
pip install 'rsn_db[mempalace]'

# Development / tests
pip install 'rsn_db[dev]'

Requires Python 3.9+.

Windows (cmd / PowerShell)

Prebuilt wheels are published for Windows (win_amd64). Install with:

pip install -U rsn_db
rsn-db --version

If pip appears stuck at "Building wheel", it is compiling Rust from source because no wheel matched your platform/Python. Either:

1. Upgrade to the latest version (needs a published Windows wheel): pip install -U rsn_db
2. Use the alias after install: rsn-db (not rsn) if another program owns that name
3. Or install Rust + Visual Studio Build Tools and wait several minutes for the source build to finish

Prebuilt wheels are published for common Linux targets; other platforms build from source via maturin (Rust toolchain required).

---

Quick start

Beginners (recommended)

from rsn_db.beginners import quick_start, insert_many, records_to_dicts
from rsn_db import Query

db = quick_start("my_app.rsndb", mode="friendly")
db.create_table("tasks", {"title": {"type": "string", "required": True}})
insert_many(db, "tasks", [{"title": "Learn RSN DB"}, {"title": "Ship the release"}])
print(records_to_dicts(db.query(Query("tasks"))))
db.save()

Context manager (auto-save on exit)

from rsn_db import open_db

with open_db("shop.rsndb") as db:
    db.create_table("products", {"name": {"type": "string", "required": True}})
    db.insert("products", {"name": "Widget", "price": 9.99})

Core API (full control)

from rsn_db import Database, Query

db = Database(
    storage_path="data.rsndb",
    encryption_key="change-me-in-production",
    compression="zstd",
    mode="professional",
)

db.create_table("users", {"name": {"type": "string", "required": True}})
db.insert("users", {"name": "Alice", "age": 30})

rows = db.query(Query("users").where_eq("name", "Alice"))
db.save()
db.snapshot("backup.rsndb")

See documentation/BEGINNERS.md for a guided walkthrough.

---

CLI

Two console commands are installed — same program, pick either:

Command	When to use
rsn	Default on macOS/Linux
rsn-db	Recommended on Windows if rsn conflicts with another tool in cmd/PowerShell

Interactive shell:

rsn
# Windows alternative:
rsn-db

Automation / agents

rsn --no-prompt -c "SHOW TABLES"
rsn-db --mode snarky -c "PULSE"
rsn-db --storage ./app.rsndb --json -c "COUNT users"
rsn-db --help

REPL commands (non-exhaustive)

Category	Examples
Tables	SHOW TABLES, DESCRIBE users, COUNT users
GraphRAG	INGEST …, GRAPH_QUERY …
Alive (Snarky)	PULSE, MOOD, VITALS, ACHIEVEMENT
MemPalace	MEMPALACE HELP, MEMPALACE SEARCH …, MEMPALACE REMEMBER …
Transactions	BATCH, COMMIT, ROLLBACK

---

MemPalace integration

RSN DB wraps the official MemPalace/mempalace package (wings → rooms → drawers, ChromaDB, local-first). Install the extra, then:

from rsn_db import open_db

with open_db("app.rsndb", mempalace=True) as db:
    db.remember("User prefers Tuesday deploys")
    print(db.palace_search("deploy schedule"))
    db.sync_to_mempalace()

Use only official sources: GitHub, PyPI, mempalaceofficial.com.

Details: documentation/mempalace.md

---

Session memory

Lightweight JSON sidecar for conversation turns — no extra install:

from rsn_db import SessionMemory

mem = SessionMemory.for_database("app.rsndb")
mem.add("user", "Remember: staging uses port 8081")
mem.add("assistant", "Noted.")
mem.save()

With MemPalace enabled, RsnDatabase.sync_to_mempalace() pushes turns into your palace.

---

GraphRAG

Ingest unstructured text and query it locally:

db.ingest("RSN DB was built with Rust and exposed to Python via PyO3.", source="docs")
print(db.graph_query("What is RSN DB built with?"))

---

Personality modes

Mode	Behavior
Professional	Minimal, neutral output
Friendly	Helpful tone with light personality
Snarky	Full commentary, 130+ remark pool, mood tracking, ambient PULSE / VITALS

Set via Python (mode="snarky"), CLI (--mode snarky), or saved preference on first run.

---

Security

RSN DB is designed with a security-first posture:

Control	Description
Encryption at rest	AES-256-GCM (optional encryption_key)
Integrity	SHA-256 checksums on persisted blocks
Path guard	Blocks absolute paths and directory traversal
DoS limits	Caps on batch size, recursion depth, command length
Safe imports	SQLite/JSON import respects declared schema types

Full write-up: documentation/security.md · documentation/threat_model.md

---

Documentation

Document	Description
BEGINNERS.md	First-time user guide
mempalace.md	Official MemPalace bridge
security.md	Security features and guidance
threat_model.md	STRIDE threat model
patchnotes.md	Version history
examples/	Usage examples

---

Development

git clone https://github.com/Enderchefcoder/RSN_DB.git
cd RSN_DB
pip install maturin pytest pytest-cov 'mempalace>=3.3.5,<4'
maturin develop --release
pytest tests/ --cov=rsn_db --cov-config=pyproject.toml
cargo clippy -- -W clippy::pedantic

---

License

MIT — see LICENSE.

---

_{Built with Rust and Python · GitHub · PyPI}