ARF-Sandbox-API / README.md
petter2025's picture
Update README.md
5582456 verified
---
title: ARF Sandbox API
emoji: 🧪
colorFrom: blue
colorTo: indigo
sdk: docker
pinned: false
license: apache-2.0
---
# ARF Sandbox API – Mock Endpoint
🧪 **This is a fully simulated, mock API.**
It does **not** use the protected Bayesian engine.
No real data is processed – all responses are deterministic examples.
Use this endpoint to test integrations and see the expected request/response format.
---
## ⚡ Rate Limit
To ensure fair usage, the API enforces a limit of **10 requests per minute per IP address**.
If you exceed this, you will receive a `429 Too Many Requests` response.
---
## 🔗 Endpoints
| Method | Endpoint | Description |
|--------|----------|-------------|
| `GET` | `/health` | Health check – returns `{"status":"ok","timestamp":...}` |
| `POST` | `/v1/evaluate` | Simulate an incident evaluation (see request/response below) |
| `GET` | `/docs` | Interactive Swagger UI (OpenAPI documentation) |
| `GET` | `/redoc` | ReDoc alternative documentation |
---
## 📦 Example Request
```bash
curl -X POST https://A-R-F-ARF-Sandbox-API.hf.space/v1/evaluate \
-H "Content-Type: application/json" \
-d '{
"service_name": "api",
"event_type": "latency",
"severity": "high",
"metrics": {"latency_ms": 450}
}'
```
### Response Model
| Field | Type | Description |
|-------|------|-------------|
| `status` | string | Always `"success"` for a valid request |
| `recommendation` | string | `"APPROVE"`, `"DENY"`, or `"ESCALATE"` |
| `risk_score` | float | Mock risk score between 0.0 and 1.0 (higher = more risk) |
| `confidence` | float | Mock confidence (0.0 – 1.0) |
| `justification` | string | Human‑readable explanation (always indicates mock) |
| `policy_violations` | list | Empty list (no real policy evaluation in sandbox) |
> 💡 **Note:** All responses are deterministic for the same input (except for the tiny pseudo‑random jitter). This makes integration tests predictable.
---
## 🔒 Real Engine
The actual ARF core engine is **access‑controlled** and available under outcome‑based pricing.
👉 [Request pilot access](https://www.arf-ai.com/signup)
Pilot customers get:
- Real Bayesian risk fusion (online + offline HMC)
- Deterministic enforcement gates
- Complete audit trail and signed intents
- Outcome‑based pricing – pay for *verified risk reduction*
---
## 🧪 Try it Live
You can test the API interactively using the built‑in Swagger UI:
**`/docs`** endpoint of this Space – or click the “App” tab above.
---
## 📚 See Also
- [ARF Official Specification](https://github.com/arf-foundation/arf-spec)
- [Public Demo UI](https://www.arf-ai.com)
- [Interactive Risk Dashboard](https://huggingface.co/spaces/A-R-F/Agentic-Reliability-Framework-v4)
---
## 🛟 Support
For questions about the sandbox or pilot access, please use the contact form on the [ARF website](https://www.arf-ai.com/contact).
---
*© ARF Foundation – Sandbox API is Apache 2.0. The real ARF engine is proprietary and access‑controlled.*