使用方法
- 复制安装命令，在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 说明输入材料与期望交付物，并对照来源检查输出。
npx skills add https://github.com/dynatrace/dynatrace-for-ai --skill dt-obs-log-semantic-mapping开始前需了解
- 安装不包含模型订阅、API 额度或访问文件的权限。
- 仓库星数和安装量反映活跃情况，不等于独立质量或安全测试。
仓库原始内容
按源仓库内容呈现，保留标题、案例、代码、表格、链接以及原文引用的演示图片。
dt-obs-log-semantic-mapping
Build and validate semantic-dictionary-aligned mappings for audit log integrations.
Purpose
Use this skill when a user wants to:
- Suggest a mapping from a raw vendor audit log payload to Dynatrace
fetch logsfields (Workflow A). - Validate a mapping against a pasted ingested log event (Workflow B1 — static).
- Validate against live tenant data via live tenant access (Workflow B2 — runtime: fetches logs by
log.source, then runs B1 on the result).
Log Classes
| Class | Description | Key namespaces | Example sources |
|---|---|---|---|
authentication | Login, logout, MFA, token | audit.*, actor.*, browser.*, device.* | CyberArk, Okta, Azure SignInLogs |
authorization | Access decisions, permission changes | audit.*, actor.*, object.* | CyberArk, Okta |
user_action | CRUD on platform resources | audit.*, actor.*, object.*, product.* | Okta, GitHub, Sonatype |
http | HTTP request/response (WAF, network devices) | http.*, url.*, server.*, geo.*, client.* | Akamai SIEM, Cloudflare |
Workflows
| Mode | Input | Source |
|---|---|---|
| Workflow A — Suggest mapping | Raw vendor log payload | references/mapping-workflow.md § Workflow A |
| Workflow B1 — Static validation | Pasted ingested log event | references/mapping-workflow.md § Workflow B1 |
| Workflow B2 — Runtime validation | log.source value + live tenant access | references/runtime-validation.md — fetches logs, then runs B1 |
Key Concepts
Content field burial: The primary validation concern. Fields in content (the raw vendor payload) that could be promoted to top-level semantic attributes but are not. The skill always inventories buried vs promoted fields and proposes OpenPipeline extraction rules to fix gaps.
Prerequisite: When proposing OpenPipeline processor extraction rules, load thedt-dql-essentialsskill first. OpenPipeline processors use DQL functions (parse,fieldsAdd,splitString, etc.) — using non-DQL syntax produces invalid rules.
Sparse mappings are valid: Integrations like GitHub or Sonatype may only populate core fields. Minimum required: timestamp, log.source, content, loglevel, audit.action, audit.identity.
References
references/data-model-notes.md— Log SD field taxonomy, audit namespace, enums, sample-derived patterns and known discrepanciesreferences/mapping-workflow.md— Intake checklist, Workflow A and B1 procedures, content field analysis, field priority orderreferences/validation-rules.md— Required fields, content/enum/type rules, discrepancy severityreferences/openpipeline-constraints.md— OpenPipeline processor command/function/operator/matcher restrictions;parseJsonunavailability +parse→fieldsFlattenalternative; iterative operators for array castingreferences/report-format.md— Mapping table, diff table, OpenPipeline sketch, Validation Summary templatesreferences/runtime-validation.md— Workflow B2: fetch live records, then run B1samples/audit-logs.json— Mapped samples: CyberArk, Okta, Azure SignInLogs, Sonatype, GitHubsamples/http-logs.json— Mapped samples: Akamai SIEM (WAF/HTTP class)- Dynatrace Log Semantic Dictionary
