LearnPrismaBrowser

A curated, offline Markdown knowledge base for Palo Alto Networks Prisma Access Browser (Prisma Browser) — built to be queried by an LLM so you can ask setup/configuration questions and get answers grounded in the official docs, each with a citation to the exact file, section, and source URL.

It covers the product end-to-end — license → activate → onboard → deploy → configure — plus identity/Entra ID integration, the routing model, and tenant/sub-tenant setup.

• ~935 documentation pages across 22 Markdown files (~11 MB)
• Every extracted page carries a *Source: link back to the live Palo Alto documentation
• Start with INDEX.md — the catalog of every document with a summary

---

What's in here

See INDEX.md for the full catalogue. In short:

Area	Where
Core product docs	prisma-access-browser-docs.md, prisma-browser-routing-model.md, the converted sec-int-… / secure-browser-for-dummies PDFs
Setup & lifecycle	prisma-browser-licensing-onboarding.md, prisma-tenants-and-sub-tenants.md
Identity / Entra ID	prisma-browser-entra-id-references.md, strata-cloud-manager-entra-id-integration.md
Dependency library	cloud-identity.md, common-services.md, prisma-access-scoped.md, strata-logging-service.md
How it was built	doc-site-to-markdown-playbook.md, scanned-pdf-to-markdown-playbook.md

---

How to use it with an LLM (the point of this repo)

This repo ships a ready-made grounded Q&A system prompt designed to stop the model inventing answers: it responds only from these files and cites the file + section + source URL for every claim. If the answer isn't in the docs, it says so instead of guessing.

The system prompt lives in LLM-PROMPT.md. Copy the text between the === BEGIN SYSTEM PROMPT === and === END SYSTEM PROMPT === markers — that block is what you paste into the LLM. Everything else in that file is usage notes, a worked example, and tuning tips.

Driving an agent? AGENT-ONBOARDING.md explains how a fresh agent should engage with this repo (orient → retrieve → read only the needed sections → cite), and WEB-AGENT-PROMPT.md is a ready-to-paste, self-contained prompt for a browsing chatbot (Option B).

Quick start

1. Copy the prompt (between the BEGIN/END markers) into your LLM's System / Custom Instructions field.
2. Give the LLM access to the documents — pick one option below.
3. Always include INDEX.md — the prompt uses it to route to the right file.
4. Ask your question. You'll get an Answer plus a Sources list.

Three ways to give the LLM the documents

Option A — Connector / RAG (recommended, most reliable). Use a tool that ingests the whole repo: ChatGPT's “Connect GitHub”, a Claude Project with this repo added, NotebookLM, or any RAG setup. It indexes all files; the prompt + INDEX.md find the right ones. Best for the full corpus — it's too big to paste.

Option B — Browsing / Git-aware model (e.g. ChatGPT with web access). The repo's raw base URL is already baked into LLM-PROMPT.md, and the prompt tells the model to bootstrap from INDEX.md automatically. So:

1. Paste the prompt (the BEGIN/END block) into the system / instructions field.
2. Ask your question — the model fetches …/main/INDEX.md, picks the relevant file(s), fetches their raw URLs, and answers with Sources.

Gotcha (common): just pasting the repo link into a chat usually makes the model read only this README and stop. Paste the prompt, not just the link, so it knows to go fetch INDEX.md and the right file. If it still doesn't fetch, tell it: “Fetch https://raw.githubusercontent.com/nialljenIT365/LearnPrismaBrowser/main/INDEX.md, choose the relevant file, fetch its raw URL, then answer with sources.” The ~6 MB strata-logging-service.md may be too big to fetch in full — for logging-heavy questions, prefer Option A.

Option C — Paste a subset. For a focused question, paste INDEX.md + the one or two relevant files (chosen from INDEX's tables) into the chat. Don't paste everything — the corpus is ~11 MB and one file alone (strata-logging-service.md) is ~6 MB.

Example

Q: Which identity providers can I use for Prisma Browser SSO, and where do I configure Entra ID?

The model answers from strata-cloud-manager-entra-id-integration.md / cloud-identity.md and lists those files + sections + the live doc URLs as its Sources — so you can click through and verify.

---

Good to know

• Grounded, with citations. The prompt requires a Sources: block on every answer (file → section → *Source: URL). A “not covered in the documentation” reply is treated as a success, not a failure — that's the anti-hallucination design. If an answer comes back with no Sources, treat that turn as failed and re-ask.
• Traceable. Every *Source: URL points to the live page on docs.paloaltonetworks.com, so any answer can be checked against the original.
• Settings. Keep temperature low (≈0–0.3) for factual retrieval.

Sources & disclaimer

All content is derived from public Palo Alto Networks documentation (crawled from docs.paloaltonetworks.com) and public product PDFs, converted to Markdown for offline/LLM use. This is an unofficial learning aid, not affiliated with or endorsed by Palo Alto Networks. For production decisions, always confirm against the official documentation linked in each page's *Source: line.