DOC-7003 fix dead and brittle links in the develop docs

[andy-stark-redis]

Second of four PRs for DOC-7003, covering Part A1, A3 and A4 — the develop/ link
fixes. Stacked on #3863, which supplies the two streamed-type anchors this PR
links to. Base is set to that branch; review or merge it first.

What changed

Fix	File
A1 — 18 anchored RESP3 links retargeted to relref	develop/programmability/lua-api.md
A3 — BLOCK_SIZE link repointed	develop/ai/search-and-query/administration/configuration.md
A4 — query string moved before the fragment	integrate/railway-redis/_index.md
A4 — MANIFESTO line anchor pinned to a SHA	develop/programmability/_index.md
A4 — boringssl line range pinned and corrected	operate/rs/security/encryption/tls/ciphers.md
A4 — two scroll-to-text fragments dropped	develop/whats-new/7-2.md

21 lines changed across 6 files. Of lua-api.md's 11 distinct anchors, 7 were dead
because the spec never had per-type headings; all 18 links now go through relref, so
the build validates the target page. The prose read "streamed aggregate data types",
and that missing "d" is exactly what made the anchor dead, so the link text was
corrected too.

One finding in the ticket is a false positive

The Wikipedia LDAP bind link is not broken, and this PR does not touch it. The audit
reported that its unescaped closing paren truncates the markdown link. The rendered HTML
disagrees:

href="https://en.wikipedia.org/wiki/Lightweight_Directory_Access_Protocol#Bind_(authenticate)"

Full href, anchor intact, no stray text after the link. CommonMark permits balanced
parentheses inside a link destination and goldmark implements that, so the detector's
regex stopping at the first paren is the bug. The target anchor Bind_(authenticate)
also still exists on Wikipedia, so the link is healthy end to end. Worth folding back
into ANCHOR_AUDIT_FINDINGS.md as a third detector bug alongside the two the audit
already documents.

Line numbers had to be verified the hard way

WebFetch could not settle either line anchor — it placed the MANIFESTO phrase on line 8
and hedged on the boringssl range with "approximately". Fetching both files at pinned
SHAs and counting lines directly gave the real answers:

• MANIFESTO — #L7 was correct all along. Pinned to 3eaa2cdc, so it stays correct.
• boringssl — var cipherSuites opens at line 99 and the list closes at line 130;
  the link said #L99-L131, one blank line past the end. Pinned to 887d5178 and
  tightened to #L99-L130.

Verification

Built the site and checked every retarget against rendered anchors, not predicted slugs:

• All 10 distinct anchors lua-api.md now targets are present as real ids on the built
  protocol spec page.
• #create-a-vector-index present on the built vectors page.
• Zero href="" on the rendered lua-api.md — no relref silently failed to resolve.
• The 18 new hrefs reconcile exactly against the per-anchor counts.

Two things to flag

A3 is a retarget under protest. BLOCK_SIZE is no longer documented anywhere on the
vectors page — the only trace is DEFAULT_BLOCK_SIZE inside a TRAINING_THRESHOLD table
cell. #create-a-vector-index is the closest honest target, but the real gap is editorial
and outlives this link fix.

Two more dead relref anchors exist on the protocol spec page, left alone as out of
scope. eval-intro.md:122 uses #resp-arrays and the generated content/commands/ft.aliaslist.md:78
uses #array; both should be #arrays. Not in Part A, and the second is regenerated
upstream so a local fix would revert. They make good test fixtures for the offline checker
in PR 3.

Also unverified: the railway #tcp-proxying anchor. This PR fixes the URL's structure —
the query string was after the fragment, so ?utm_medium=… was being parsed as part of the
anchor — which is correct regardless. Whether that anchor exists rests on one weak negative
fetch, so I have not claimed it either way.

🤖 Generated with Claude Code

---

Note

Low Risk
Documentation-only link and anchor updates with no changes to application code, configuration defaults, or security-sensitive runtime behavior.

Overview
This PR repairs broken or fragile documentation links across six files as part of DOC-7003 (develop link fixes), with no product or API behavior changes.

The largest edit is in lua-api.md: 18 external RESP3 GitHub spec URLs are replaced with relref links to protocol-spec anchors (#maps, #sets, #nulls, #booleans, and related types). Unsupported Lua API types are linked to #attributes, #streamed-strings, and #streamed-aggregated-data-types, and the prose is corrected from “streamed aggregate” to “streamed aggregated” to match the real heading.

configuration.md repoints the BLOCK_SIZE mention from a dead vectors anchor to #create-a-vector-index. programmability/_index.md pins the Redis MANIFESTO GitHub link to a commit SHA instead of unstable. ciphers.md pins the BoringSSL cipher list link to a SHA and tightens the line range to L99–L130.

railway-redis/_index.md moves UTM query parameters before the #tcp-proxying fragment so tracking params are not treated as part of the anchor. whats-new/7-2.md drops brittle scroll-to-text fragments from two redis.io blog URLs.

^{Reviewed by Cursor Bugbot for commit cfd0c87. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

DOC-7003

[github-actions]

Staging links:
https://redis.io/docs/staging/DOC-7003-develop-link-fixes/develop/ai/search-and-query/administration/configuration/
https://redis.io/docs/staging/DOC-7003-develop-link-fixes/develop/programmability/
https://redis.io/docs/staging/DOC-7003-develop-link-fixes/develop/programmability/lua-api/
https://redis.io/docs/staging/DOC-7003-develop-link-fixes/develop/whats-new/7-2/
https://redis.io/docs/staging/DOC-7003-develop-link-fixes/integrate/railway-redis/
https://redis.io/docs/staging/DOC-7003-develop-link-fixes/operate/rs/security/encryption/tls/ciphers/

[github-actions]

🧠 Redis Memory

Found 8 related items from repository history (3 new this commit):

• PR — DOC-7003 document RESP3 streamed types in the protocol spec

  First of four PRs for DOC-7003. Adds two sections to the protocol spec page so the

• PR — DOC-6909 Fix pre-existing broken links across the docs

  Part 1 of 4 of a stack that splits DOC-6909 Add link render hook to replace relref (investigation) #3732 into independently reviewable pieces, after @paoloredis noted the single PR was too much to review

• PR — DOC-6909 Teach the feed pipeline and version archiver the /content/ link form

  Part 2 of 4. Stack: DOC-6909 Fix pre-existing broken links across the docs #3795 → DOC-6909 Teach the feed pipeline and version archiver the /content/ link form #3796 (this) → DOC-6909 Add link and blockquote render hooks to replace relref #3797 → DOC-6909 Migrate develop/clients/redis-py to portable Markdown links and callouts #3798

• PR — DOC-6955 Repoint the Google ADK links at Agent Memory's new home

  Fixes DOC-6955. Ten relref calls across three Google ADK pages named Agent Memory paths

• Commit — DOC-6909 Fix broken links surfaced by the render-hook investigation

  DOC-6909 Fix broken links surfaced by the render-hook investigation

• Commit — DOC-7003 fix dead and brittle links in the develop docs 🆕

  DOC-7003 fix dead and brittle links in the develop docs

• PR — DOC-7003 retarget dead k8s API anchors to the local reference 🆕

  Third PR for DOC-7003, covering Part A2 — the Kubernetes API anchor retargets.

• Commit — DOC-7003 retarget dead k8s API anchors to the local reference 🆕

  DOC-7003 retarget dead k8s API anchors to the local reference

Memory updated at cfd0c87

[dwdougherty]

Thank you for cleaning these up! Approved.

[andy-stark-redis]

Thank you for cleaning these up! Approved.

No problem, @dwdougherty - thanks for the review :-)