HDDS-15526. Add security threat model (THREAT_MODEL.md + SECURITY.md + AGENTS.md)

AGENTS.md

@@ -195,3 +195,33 @@ Key paths:
 - [`hadoop-hdds/dev-support/checkstyle/checkstyle.xml`](./hadoop-hdds/dev-support/checkstyle/checkstyle.xml)
 - [`dev-support/rat/rat-exclusions.txt`](./dev-support/rat/rat-exclusions.txt)
 - [Ozone Enhancement Proposals](https://ozone.apache.org/docs/edge/design/ozone-enhancement-proposals.html)
+
+## Security
+
+When assessing a potential security vulnerability in Apache Ozone, complete
+these steps before drafting any report or reaching any security conclusion.
+
+### Step 1 — Read the threat model
+Read **[THREAT_MODEL.md](THREAT_MODEL.md)**: the multi-service trust boundaries,
+the **secure mode** knob, the properties provided vs. left to the operator, and
+the known non-findings.
+
+### Step 2 — Read the security policy
+Read **[SECURITY.md](SECURITY.md)** for how to report.
+
+### Key scoping facts (see THREAT_MODEL.md)
+- Ozone is a cluster of network services (S3 Gateway, OM, SCM/internal-CA,
+  Datanodes/Ratis, Recon). Roles: untrusted client, authenticated-but-
+  unauthorized user, operator, service peer, bounded-Byzantine datanode.
+- **Secure mode** (`ozone.security.enabled=true`) is load-bearing: a finding
+  that only manifests in non-secure (dev) mode is out of model (section 5a).
+- Ozone does **not** own its dependencies' security — the Kerberos KDC, Ranger
+  policy correctness, the SCM CA private key, KMS keys, and network isolation
+  are the operator's (sections 3/9/10). Route such findings there.
+- Ratis (Raft) safety holds under an honest majority; a Byzantine majority is
+  out of scope.
+- integration-test modules, and test utilities are out of scope.
+
+### Then assess
+Route the finding to exactly one disposition in **THREAT_MODEL.md section 13**,
+citing the section. If it cannot be routed, it is a `MODEL-GAP` — surface it.

SECURITY.md

@@ -13,3 +13,13 @@ This email address is a private mailing list for discussion of potential securit
 This mailing list is **NOT** for end-user questions and discussion on security. Please use the dev@ozone.apache.org list for such issues.
 
 In order to post to the list, it is **NOT** necessary to first subscribe to it.
+
+## Threat Model
+
+A threat model for Apache Ozone is maintained in [THREAT_MODEL.md](THREAT_MODEL.md).
+It describes the multi-service trust boundaries (S3 Gateway, OM, SCM/CA,
+Datanodes/Ratis), the load-bearing role of **secure mode**
+(`ozone.security.enabled`), the properties Ozone provides versus those left to
+the operator (Kerberos KDC, Ranger policy correctness, SCM CA key, KMS, network
+isolation), and the recurring non-findings. Triagers of scanner, fuzzer, or
+AI-generated findings should route them through `THREAT_MODEL.md` section 13.