Reference Libraries

Fourteen verified indexes of primary sources — one per major topic. Where the rest of this book explains concepts, these pages tell you which document to open for the authoritative answer, and in what order to read things.

Each entry records, where it exists:

• Official documentation and the developer / API portal
• The source repository
• SDKs and notable related repos
• Downloadable or offline documentation (PDFs, doc tarballs, offline bundles)
• A short honest note — what the source is actually good for, and where it misleads

Each page then has a two-track Education section (Basic and Advanced, with university course material in both), a Research papers & open-access literature section covering free paper sources for that field, a Video courses, channels & talks section of verified channels and lecture playlists, and a Conference videos, notes & archives section covering the conference channels, video archives, proceedings portals and paper-adjacent note archives for that field. The DSA index is the exception in placement only: its video material sits in its Community & video category and in the Basic education track rather than in a separate section.

The indexes

1,257 entries, 2,744 unique URLs across the fourteen indexes (entries counted per index; the URL total is de-duplicated across all fourteen pages and was recomputed on 2026-10-09 — the previous figure did not reconcile with the per-index column).

Research papers

Every index except the networking one ends with a research section. Fifteen sources are shared across all nine — arXiv (with its API and bulk-data endpoints), ar5iv, alphaXiv, Semantic Scholar, OpenAlex, DBLP, OpenReview, CORE, Unpaywall, Papers We Love, The Morning Paper archive, USENIX Proceedings, the ACM DL, IEEE Xplore and DROPS/LIPIcs — and each page adds the venues specific to its field. (The networking index instead carries RFC/standards bodies, NOG communities and operator resources in its own categories.)

The access model is recorded for every venue: which are fully open (USENIX, PVLDB, PMLR, JMLR, ACL Anthology, CVF, LIPIcs, PACMPL), which are partially open (ACM), and which are mostly paywalled with a legal free route around them (IEEE Xplore → arXiv, author pages, Unpaywall).

Verification method

Every URL was fetched with redirects followed and a 25-second timeout, and recorded with its final status and resolved URL.

• Dead links were replaced, not kept. Where a documented path 404’d, the working replacement is used.
• Bot-blocked sources are flagged, not dropped. Several important sources (the Intel SDM, developer.arm.com, the OSDev Wiki, dev.mysql.com, cppreference, the ACM DL) return 403 to automated clients while working normally in a browser. These are kept with an explicit note, because dropping them would make the indexes worse.
• Ownership and naming changes are noted inline where they affect whether a link or a search will work: Linode → Akamai Cloud Computing, Timescale → TigerData, Redis → the Valkey fork, Terraform → OpenTofu, Coq → Rocq, Talos docs → docs.siderolabs.com, Spirent → absorbed into Keysight, OpenZiti docs → hosted under NetFoundry, NS1 → IBM NS1 Connect.
• Two sources are kept with a browser-only flag in the newest indexes: ai.google.dev redirects automated clients to a Google sign-in page, and openai.com/research/ returns 403 to them. Both load normally in a browser.

Machine-readable data

Each index also ships as CSV, with the same columns and one row per resource: every entry, every education resource, every video resource and every conference source. Category values are the numbered entry categories (3. Competitive math), Education — Basic, Education — Advanced, Video — <group> and Conference — <group>. The DSA index keeps the original un-numbered category names; networking.csv keeps its older Vendor / Project column header.

Category, Name, Documentation, Developer / API portal, Source / GitHub, SDKs & notable repos, Downloadable / offline docs, Notes

• networking.csv
• computer-architecture.csv
• operating-systems.csv
• database-systems.csv
• distributed-systems.csv
• machine-learning-ai.csv
• languages-compilers.csv
• cloud-computing.csv
• prompt-engineering.csv
• agentic-engineering.csv
• security.csv
• concurrency.csv
• storage.csv
• dsa-competitive-programming.csv

How to use these pages

1. Starting a topic? Read that page’s Education → Basic track in order. It is deliberately short and opinionated.
2. Need the authoritative answer? Go to the entry for the system and open its Docs link, not a blog post.
3. Going deep? The Advanced track points at the source you should be reading — teaching kernels before production kernels, LevelDB before RocksDB, nanoGPT before transformers.
4. Writing something original? Start from the research section and the access notes, so you spend your time reading rather than hunting for PDFs.

Every page ends with an “If you only do three things” box and an honest notes section covering the traps — which documentation is genuinely excellent, which is stale, and which project renamed itself out from under its own links.

Driving these indexes with a model

Documentation Navigator — Generic Prompt Library is a set of 16 topic-agnostic prompts built for exactly this corpus. Set {{TOPIC}}, attach the relevant index, and the model answers by pointing at documents and a reading order instead of improvising prose.