Progress tracker¶

Where the playbook stands. Updated as work lands. Last update: 2026-11-14.

Legend: ✅ done · 🟡 in progress · ⬜ not started

Phase 1: content ✅¶

#	Task	Status	Notes
1	Scaffold, licenses, CI, note checker	✅	`scripts/check_notes.py` (16 unit tests) and runnable-snippet tests (8 snippets)
2	Note template, style guide, glossary	✅	70 glossary terms
3	00 · Foundations	✅	6 notes
4	01 · Coding agents + P1	✅	8 notes; P1: 20 tests
5	02 · Building agents	✅	6 notes
6	04 · LLM security + P2	✅	13 notes (OWASP LLM 2026, agentic, incidents, frameworks); P2: 26 tests + 5 expected failures
7	03 · AI safety and guardrails	✅	5 notes + defense-layers map
8	05 · Evaluation, 06 · Memory + P3	✅	8 notes; P3: 11 tests, eval 6/6
9	P4 · Guardrailed support agent	✅	33 tests, eval 12/12
10	07 · Agent ops, 08 · System design + P5	✅	10 notes incl. 3 case studies; P5: 5 tests, Docker + Jaeger verified, design doc, threat model
11	09 · Careers, glossary, README, START-HERE, ABOUT	✅	4 career notes
12	All checks green; your review of the private repo	✅	60 notes pass the checker; 0 markdown lint issues; 304 unique links OK; all project tests pass in CI. Reviewed and published

#	Task	Status
13	MkDocs Material site built from the same markdown	✅
14	Visual polish (brand colors, logo, home page cards)	✅
15	Make public and deploy GitHub Pages	✅ Live at https://neuralsaasai.github.io/ai-engineering-playbook/
16	Move to a NeuralSaas AI domain (needs your yes and a DNS change)	⬜ Planned for later

Built with MkDocs Material from the repo root (mkdocs.yml, site_hooks.py, overrides/, assets/). The notes are unchanged.
Navigation is generated from each module README's table, so adding a note to a module README adds it to the site.
Links to code and data files point to GitHub automatically.
Logo: assets/logo.svg (icon), assets/logo-wordmark.svg and assets/logo-wordmark-dark.svg, PNG icons for browsers and phones.
SEO: descriptive page titles, meta descriptions from each note's first section, canonical URLs, sitemap.xml with each note's review date, robots.txt, Open Graph and Twitter cards, schema.org structured data (Course on the home page; TechArticle/LearningResource and breadcrumbs on notes), web app manifest, and llms.txt for AI assistants.
GitHub discoverability: keyword-rich description, 20 topics, homepage link, CITATION.cff ("Cite this repository").
The site URL comes from the SITE_URL repository variable, so the custom-domain move is a config change.
Preview locally: make site-serve.
Watch item: MkDocs 1.x is no longer actively maintained, and the Material team points to its successor (Zensical). The build is pinned and works today. Re-evaluate in 2027.

Item	Command
Upload the GitHub social preview (Settings > General > Social preview)	Use `assets/social-card.png`
Register the site in Google Search Console and submit the sitemap	Add the HTML-tag token as repo variable `GOOGLE_SITE_VERIFICATION`, re-run the Website workflow, then submit `sitemap.xml`
Fewer permission prompts in Kiro	`mkdir -p ~/.kiro/settings && cp .local/kiro-permissions.yaml ~/.kiro/settings/permissions.yaml`

Projects use deliberately simple, regex-based guardrails and fake models so they run offline. Each README suggests swapping in real libraries and models as an exercise.
P5's rate limiter keeps counters in memory, so the container runs one worker. Verified: with two workers each process counted separately and no requests were limited.
Prices and plan names in the tool comparison are as of 14 Nov 2026 and need a monthly re-check.

Date	Decision
2026-11-14	Repo name `ai-engineering-playbook`; notes + 5 full projects; no reference to source video or roadmap.sh
2026-11-14	Default live model: Gemini free tier via LiteLLM; tests use fake models, no API keys in CI
2026-11-14	Security notes follow the OWASP LLM Top 10 2026 (published Aug 2026), with 2025 IDs shown alongside
2026-11-14	Repo made public; commit history attributed to jindalAnuj (GitHub noreply email)
2026-11-14	Domain: start on github.io, then `learn.neuralsaasai.com` or `neuralsaasai.com/playbook`