Domain Context

---

name: domain-context description: >- Chapter-indexed, self-contained book knowledge base and domain specialist guidance for azure-devops-cicd. Use the machine-readable retrieval index to find comprehensive chapter guides, examples, tradeoffs, and decision rules across twenty-five books and twelve engineering domains without requiring local EPUB/PDF files. Current official product documentation still wins.

Domain Context

Purpose

Give an agent enough durable domain knowledge to reason independently about an engineering problem before it recommends an Azure DevOps implementation. The packaged book references are chapter-complete guides, not short checklists. They preserve the books' important mental models, mechanics, examples, tradeoffs, failure modes, and decision criteria in original, paraphrased language.

Books remain design input. They are not evidence of current Azure DevOps, Azure, Kubernetes, OpenAI, language, framework, or tool behavior.

Claude hosts may use agents/domains/*.md as a fast application lens. Other hosts should use references/domains/<domain>.md. Both routes lead to the same chapter-indexed book knowledge base.

Retrieval contract

Do not load all twenty-five book files for every task. Retrieve in two stages:

1. Translate the request into likely domains, concepts, products, acronyms, examples, and alternate search terms.
2. Search references/chapter-index.json. Its chapters records include summary, topics, concepts, technologies, examples, search_terms, and related_domains.
3. Select the smallest sufficient set of chapters, normally one to four. Use the record's path and anchor to open the exact chapter section.
4. Open references/domains/<domain>.md when a deployment or review checklist is useful. The checklist is an application lens, not a substitute for the chapter guidance.
5. Read adjacent chapters when the selected chapter depends on a larger lifecycle, architecture, or operating model.
6. Verify every time-sensitive product claim against current first-party documentation before presenting it as fact.
7. In the answer, distinguish source-backed book patterns, observed customer evidence, current product facts, and your own judgment.

The generated human catalog references/chapter-index.md is useful when JSON search is unavailable.

Retrieval examples

• A request about HPA stabilization, KEDA, or scale-to-zero should search chapter metadata for those terms, then open the matching autoscaling chapters rather than every Kubernetes book.
• A request about pipeline templates, build once, or environment approvals should search the CI/CD books, then use the cicd-design-patterns domain checklist to translate the book concepts into an assessment.
• A request about agent handoffs, guardrails, or RAG evaluation should search the AI-agent chapter records and read the specific design and evaluation chapters before proposing pipeline gates.

Packaged knowledge base

The plugin is self-contained. EPUB and PDF binaries are not committed and no runtime path under ~/Downloads is required.

skills/domain-context/references/
├── book-index.md             # ISBN, stub, domain, and book path
├── chapter-index.json        # machine-readable retrieval metadata
├── chapter-index.md          # generated human chapter catalog
├── books/<title-kebab>.md    # comprehensive chapter guides
├── domains.md                # compact cross-domain routing
└── domains/<domain>.md       # domain-specific application checklists

Each book file contains:

• hidden book-meta JSON with canonical identity and book-level tags;
• one hidden chapter-meta JSON record per numbered content chapter;
• a stable chapter anchor such as 2491-ch04;
• detailed, paraphrased chapter guidance and book-derived examples;
• cross-cutting synthesis, decision rules, and an agent retrieval checklist;
• original Mermaid diagrams when a relationship is materially clearer visually.

The metadata indexes are generated from the book Markdown source by scripts/generate_book_kb_index.py. Do not hand-edit generated indexes.

Domain routes

Domain	Application lens	Comprehensive book references
cicd-design-patterns	domain	CI/CD Design Patterns, Implementing CI/CD Using Azure Pipelines
powershell-automation	domain	Mastering PowerShell Scripting
gitops-delivery	domain	Implementing GitOps with Kubernetes
azure-cloud-native	domain	The Azure Cloud Native Architecture Mapbook
kubernetes-platform	domain	The Kubernetes Bible, Kubernetes - An Enterprise Guide, 50 Kubernetes Concepts, Big Data on Kubernetes, Kubernetes Autoscaling
dotnet-engineering	domain	Clean Architecture with .NET, Real-World Web Development with .NET 10, C# 14 and .NET 10 Fundamentals
typescript-clean-code	domain	Clean Code with TypeScript
python-ai-apps	domain	Building AI Intensive Python Applications
ai-agent-engineering	domain	30 Agents Every AI Engineer Must Build, Building Agents with OpenAI Agents SDK, Building AI Agents with LLMs, RAG, and Knowledge Graphs, The AI Product Manager's Handbook
observability-finops	domain	Observability in the AI-Native Era, Efficient Cloud FinOps, Platform Engineering for Architects
enterprise-api	domain	Enterprise API Management
solutions-tpm	domain	Solutions Architect's Handbook, Technical Program Manager's Handbook

The canonical ISBN and stub mapping is in references/book-index.md.

Evidence and citation rules

Use this hierarchy:

1. live, read-only customer or repository evidence;
2. current first-party product documentation;
3. customer-owned standards and accepted architecture decisions;
4. chapter-specific book patterns from this knowledge base;
5. explicitly labeled professional judgment.

Apply these rules:

1. Current first-party documentation wins when it conflicts with a book.
2. Label book-derived advice as Book pattern (<stub>, ch. <number>) when a specific chapter materially supports it.
3. Cite the book title or stub and chapter, not the local EPUB/PDF path.
4. Do not represent a paraphrased book example as observed customer evidence.
5. Do not paste long excerpts or large copied code listings into Wiki pages, PR comments, commits, or generated deliverables.
6. Use book examples to explain a mechanism or tradeoff, then adapt the decision to the actual environment.
7. Do not use em dashes in Azure DevOps-facing output that cites books.

Visual and copyright policy

The public plugin does not redistribute proprietary book figures. When a visual materially improves understanding, the book reference uses an original Mermaid diagram that expresses the relationship in new form. Treat those diagrams as explanatory models, not replicas of publisher artwork.

Agent response checklist

Before returning book-backed guidance:

• Did you search chapter metadata before opening long references?
• Did you load the exact chapter sections that support the recommendation?
• Did you preserve the difference between a book example and customer evidence?
• Did you state relevant tradeoffs and failure modes, not only a preferred pattern?
• Did you verify time-sensitive product behavior?
• Did you cite the stub and chapter?
• Did you avoid requiring local source books at runtime?

Non-goals

• Shipping EPUB/PDF binaries or proprietary figures
• Reconstructing books verbatim
• Treating book guidance as current product documentation
• Loading the entire book corpus for every request
• Using checklists without reading relevant chapter context