FAQ¶

Frequently asked questions about the scope, selection process, and maintenance of this list. If your question is not answered here, open a Q&A discussion.

Scope & Selection¶

Why isn't tool X listed?¶

Every entry must meet the four Quality Standards:

Production focus — it solves a real-world problem like scalability, latency, or reliability.
Activity — the project has been updated within the last 6 months.
Documentation — a clear README, API reference, or case study exists.
No "marketing only" links — open-source tools and transparent technical posts are preferred over marketing fluff.

A tool that does not meet all four is not listed, no matter how popular it is. If you believe a tool qualifies, see How do I propose a new tool?

Why aren't Dify, Flowise, Open WebUI, or AnythingLLM listed?¶

They are excellent projects, but they are end-user applications and low-code platforms, while this list focuses on infrastructure building blocks for engineering teams that assemble and operate their own production RAG pipelines — ingestion, retrieval, evaluation, observability, serving, and so on. The weekly discovery feed regularly surfaces these projects as raw candidates, and they are triaged against this scope rule.

Which framework or vector database is "best"?¶

The list deliberately avoids absolute rankings — the answer depends on your constraints. Start with the Decision Guide, the per-category comparison tables, and the Reference Architectures. Numeric comparisons live in benchmarks.md with explicit sources and evidence tags.

Contributing¶

How do I propose a new tool?¶

Read CONTRIBUTING.md, then either open an issue with the Add Resource template or submit a PR that follows the two-line bullet format and alphabetical order. Explain the production problem the tool solves — that is the main acceptance criterion.

Can I submit my own project?¶

Yes. Self-submissions are welcome and are evaluated against exactly the same Quality Standards as any other entry. Please disclose your affiliation in the issue or PR description.

Why was my issue or PR marked stale?¶

Issues and PRs with no activity for 60 days are labeled stale and closed 7 days later by automation. Leaving any comment resets the timer. If your contribution was closed and is still relevant, feel free to reopen it or open a new one.

Evidence & Benchmarks¶

What do the [3P], [V], and [A] tags mean?¶

Every numeric claim in this repository carries one of three evidence tags:

[3P] — third-party measured (academic paper, independent benchmark, neutral reproduction).
[V] — vendor-stated (the vendor's own blog, docs, or whitepaper).
[A] — anecdotal (self-reported production case study, engineering talk).

The full policy is described in the Evidence Tier section of CONTRIBUTING.md.

Why was a number I added removed or hedged?¶

Numeric claims must supply all four Evidence Tier fields: source URL, date, tag, and (when available) a methodology link. Claims that cannot supply them are moved to benchmarks.md § Gaps or removed entirely. We prefer 20 well-cited rows over 80 half-cited ones.

Maintenance & Style¶

When are tools removed from the list?¶

Per the Removal & Deprecation Policy: inactivity for 6+ months, an archived or deleted upstream repository, a clearly superior replacement, or retracted evidence. The weekly discovery engine automatically flags listed repositories with no push in 180+ days. Widely referenced tools are soft-deprecated (annotated inline) rather than deleted; hard removal is reserved for dead links and archived repos.

Why does this list deviate from strict awesome-lint?¶

Deliberately. The two-line bullet format, decision guides, reference architectures, and evidence tables are the production-grade differentiators of this list, and they deviate from strict awesome-lint formatting. The linter runs as advisory (non-blocking) CI for that reason — see the style notes in CONTRIBUTING.md.

Is the list available in languages other than English?¶

No — English only, by design. A single source language keeps every entry, benchmark citation, and policy document in one reviewable place; translations drift out of date quickly in a list that changes weekly. Translation PRs are therefore not accepted.