jhansi 0.1.7

Python SDK for jhansi.io - run AI-generated code in isolated sandboxes

jhansi.io

Build it. Run it. Ship it.

Run AI-generated code in isolated sandboxes. Open source. Self-hostable.

---

The problem

Every AI coding tool generates code. None of them answer:

"Where does that code actually run, and what can it touch?"

Paste it into your terminal and hope? That's eval() with access to your .env, your AWS creds, your customer database.

In a startup that's risky. In fintech that's an FCA conversation you do not want to have.

jhansi.io is the answer.

---

What it is

A sandboxed execution platform with three layers:

Layer	Status	What it does
jhansi SDK	🟢 Available	Python SDK — pip install jhansi
Petri	🟢 Available	Execution engine — Docker isolation, multi-language, self-hostable
TenantVault	🔜 Coming soon	Per-tenant secrets injection, envelope encryption, audit log

---

Quick start

# Start Petri
git clone https://github.com/jhansi-io/petri.git
cd petri
docker compose up

# Install the SDK
pip install jhansi

from jhansi import Sandbox

with Sandbox(language="python") as sb:
    sb.upload_file("main.py")
    result = sb.exec("python main.py")
    print(result["output"])

Full docs at docs.jhansi.io.

---

Repositories

Repo	Description
jhansi-io/jhansi	SDK — pip install jhansi
jhansi-io/petri	Execution engine — FastAPI, Docker isolation, multi-language

---

Why jhansi.io?

• Competitors (E2B, Modal, Daytona) are SaaS only — jhansi.io is open core, self-hostable
• Nobody else combines execution sandbox + secrets vault + compliance audit in one platform
• Built by someone with 18 years in banking who has lived the compliance problem firsthand
• Regulation (FCA, SOC2, EU AI Act) requires audit of AI actions — jhansi.io provides it

---

Roadmap

Full roadmap: jhansiio.featurebase.app

---

Follow the build

Building in public. Technical deep-dives on Dev.to.

Star this repo to follow along. ⭐

---

License

Apache 2.0 — see LICENSE.