> ## 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.

# Per-Sentinel Security Reports

> Milestone 2 per-Sentinel security reports: aggregate attack summaries, security scores, evidence blob indexes, and wallet-free shareable verification links, stored as immutable Walrus blobs.

# Per-Sentinel Security Reports

*Milestone 2 deliverable. This page documents the report-generation pipeline and report schema —
the public report surface ships with the Milestone 2 release.*

Per-Sentinel Security Reports aggregate a Sentinel's entire attack history into a single immutable
Walrus blob: a structured, shareable security report that a defender can hand to customers,
auditors, or a community as proof of their Sentinel's track record.

## Report contents

Each report blob contains:

| Section | Content |
| - | - |
| **Attack summary** | All attacks attempted against the Sentinel: total attempts, outcomes, and the distribution of attack classifications (prompt injection, jailbreak, data exfiltration, …). |
| **Security score** | The Sentinel's aggregate security score — a resistance metric derived from jury scores and breach outcomes across all recorded attacks. |
| **Evidence index** | The Walrus blob ids of **all** underlying evidence blobs associated with the Sentinel, so every claim in the report resolves to a verifiable primary record. |
| **Verification link** | A shareable, verifiable URL served from the [Public Evidence Explorer](https://docs.redsentinel.xyz/walrus/evidence-explorer.md). Recipients need **no Sui wallet and no Sui infrastructure** to open it. |

## Report schema (sketch)

```json theme={null}
{
  "reportVersion": 1,
  "sentinelId": 42,
  "chain": "sui",
  "generatedAt": "2026-10-14T00:00:00.000Z",
  "attackSummary": {
    "totalAttacks": 1284,
    "breaches": 9,
    "repelled": 1275,
    "classifications": { "prompt_injection": 640, "jailbreak": 401, "…": 243 }
  },
  "securityScore": { "value": 0.97, "method": "resistance-v1" },
  "evidenceBlobIds": ["…", "…"],
  "reportBlobId": "…"
}
```

## Generation pipeline

1. **Collect.** On a schedule, the report worker gathers all certified evidence blob ids and
   verdict metadata for the Sentinel from the platform's anchoring records.
2. **Aggregate.** Attack outcomes and classifications are rolled up; the aggregate security score
   is computed from the jury scores in the underlying evidence records.
3. **Store.** The report JSON is written to Walrus as an immutable, non-deletable blob (same
   storage path as evidence records — [Evidence Pipeline](https://docs.redsentinel.xyz/walrus/evidence-pipeline.md)).
4. **Anchor.** The report blob id is anchored on-chain next to the Sentinel's identity, and the
   report becomes reachable at a stable shareable URL.

Because reports are derived only from **certified evidence blobs** (never from mutable platform
state), anyone can recompute the aggregates from the evidence index and check the report — the
report is a convenience view over verifiable primary records.

## Status

Ships with the Milestone 2 release. The [Milestone 2 Status Report](https://docs.redsentinel.xyz/walrus/milestone-2-status.md)
tracks the deliverable KPIs (≥ 50 aggregate reports generated and stored).


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