> ## Documentation Index
> Fetch the complete documentation index at: https://docs.redsentinel.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Data Access Architecture

> The access control architecture for Red Sentinel Walrus data — raw evidence, public aggregated insights, and subscriber-only raw feeds.

# Data Access Architecture

Red Sentinel's Walrus integration produces data at different sensitivity levels. The access
architecture separates **raw data access for paying subscribers** from **aggregated insight access
for the public**, while keeping everything that backs a public claim itself public and verifiable.

## Access tiers

| Tier | Who | What | How it's enforced |
| - | - | - | - |
| **Raw evidence blobs** | Public | Individual attack evidence records, readable by blob id from any Walrus aggregator. | Walrus itself — blobs are public by design; a reader needs the blob id. |
| **Aggregated insights** | Public | [Per-Sentinel Security Reports](https://docs.redsentinel.xyz/walrus/security-reports.md) and the [Public Attack Dataset](https://docs.redsentinel.xyz/walrus/attack-dataset.md), plus explorer feeds and counters. | Public web surfaces + unauthenticated API endpoints; reports/datasets are Walrus blobs anchored on-chain. |
| **Raw data feeds (subscribers)** | Paying subscribers | Bulk/export access to raw attack records, non-anonymized evaluation internals, and early dataset drops. | Backend subscription gating on API routes; credentials and entitlement checks at the API boundary. |

## Principles

1. **Public claims have public proof.** Anything surfaced publicly (a report score, a KPI counter,
   a shared record link) derives from data that is itself publicly retrievable — certified Walrus
   blobs and on-chain events. Aggregation never promotes private data into a public claim.
2. **Anonymization happens before publication, not after.** The public dataset tier only ever sees
   scrubbed, de-identified records (see [Attack Dataset](https://docs.redsentinel.xyz/walrus/attack-dataset.md#anonymization-approach));
   identity-bearing fields never enter public blobs.
3. **Gating lives at the API, not at the storage layer.** Raw feeds for subscribers are gated by
   the platform API (authentication + entitlement checks). Walrus blobs, once public, cannot be
   un-published — which is precisely why identity-bearing data is excluded from public blobs at
   write time.
4. **Wallet-free reads for public tiers.** Verifying a report or an evidence record requires no
   wallet, no tokens, and no Sui infrastructure — see the
   [Public Evidence Explorer](https://docs.redsentinel.xyz/walrus/evidence-explorer.md).

## Data flow summary

```
 attack record ──► evidence blob (public, Walrus)  ──► explorer / permalinks
        │                                                 ▲
        ├──► aggregated reports (public, Walrus) ─────────┘
        ├──► anonymized dataset (public, Walrus, on-chain anchored)
        └──► raw internals (private store) ──► subscriber feeds (API-gated)
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.