ADORSYS-GIS / ADORSYS-GIS/lightbridge-authz

[Epic]: SIEM-exportable audit log

未关闭
#257 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
epic
主要语言
Rust
星标
0
派生
1
平均合并
7 小时 7 分钟
30 天内合并 PR
237

描述

## Summary
Ship a structured, SIEM-exportable audit log covering policy changes, model access, admin actions and credential lifecycle — not just application request logs — with a documented retention policy and tamper-evidence.

## Intent / Source of truth
A CISO will not approve a tool their SOC cannot see into. GitHub's enterprise bar is a 180-day-retention audit log streamable to a SIEM. Source: 2026-08-13 enterprise-readiness research pass; no SIEM-exportable audit surface exists today.

## Scope
- [ ] Structured audit event schema covering policy/access/admin/credential events
- [ ] Streaming export to Splunk, Microsoft Sentinel, and a generic webhook sink
- [ ] Documented retention policy (minimum 180 days, configurable)
- [ ] Tamper-evidence mechanism

## Out of scope
- Real-time SIEM correlation rules (that's the customer's SOC tooling, not ours)
- Historical backfill of events predating this epic

## Verification
A policy change, a model-access event, an admin action and a credential-lifecycle event each produce a correctly schema'd audit record; the record is visible in a test Splunk/Sentinel instance within the documented export latency; a tampered record is detectably invalid.

## Risk assessment
An audit log that is itself mutable by admins defeats its purpose for compliance; tamper-evidence must cover admin-level access, not just external attackers.

## AI Usage Declaration
Drafted with AI assistance during the 2026-08-13 cross-repo backlog consolidation and enterprise-readiness research. A human owns intent, verification and consequences.

贡献指南

打开贡献指南

调研方向

No files, tests, or code entry points are named. Start by mapping the existing audit-event and export surfaces in the Rust auth service, then define the schema, retention, three export targets, and tamper-evidence boundaries from the stated scope. Done means the four event categories pass verification, exports reach the specified sinks within documented latency, and tampering is detectable.

由索引模型根据 Issue 内容生成。

评估

技术栈
rust
领域
observability, security
Issue 类型
功能
难度
5/5
预计耗时
一周以上
活跃度
冷清
描述清晰度
基本清楚
新手友好度
30/100

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。