[Programcode]

Spec 2, part 2 of 2. Stacked on #233. Merge #233 first. This PR targets main, so until #233 merges, GitHub also shows #233's commits here. This PR's own changes are the diff seo/deepen-thin-guides...seo/merge-thin-guides.

What changed

5 weak pages merged into their survivors. Each gets a 301 in vercel.json and an entry in retiredGuides in src/test/vercel-deployment-config.test.ts. Its PAGES and ANSWERS entries and its public/<slug>/ folder are removed, and its best material is folded into the survivor:

Retired	301 to	What the survivor gained
/debate-collapsing-strategy	/debate-clash-points	Now "Clash and collapse", with a section on where each format collapses
/debate-prep-time-management	/how-to-prepare-for-a-debate-tournament	A prep-time-by-format table and a section on spending prep in round
/debate-block-file-guide	/how-to-practice-rebuttals	A "Build a block file" section with an example block
/debate-pronunciation-and-delivery	/debate-speech-structure	Now "speech structure and delivery", with a delivery section
/ai-vs-traditional-debate-prep	/ai-debate-practice	An "AI vs traditional prep" section with a task table

Two tool pages linked to retired slugs: the timer page and the speech calculator page. Both now link to the survivors. scripts/gsc_submit.py lists none of the retired URLs. It now includes the 3 survivors it was missing.

6 skills pages deepened. These are clash points, tournament prep, speech structure, fallacies, motion analysis and judge adaptation. Each one has:

• a worked example on an original practice motion
• a table
• "Common mistakes" and a 12 to 15-minute drill
• 4 FAQs of its own
• a rewritten lead answer that the sections don't repeat

Prep times and speech times come from src/lib/free-tools/formats.ts. Primary sources I opened:

• the NSDA High School Unified Manual and the WUDC manual, on tournament prep
• the WUDC manual's "squirrel" rule, on motion analysis
• the WUDC manual's "ordinary intelligent voter" standard, on judge adaptation
• the Stanford Encyclopedia of Philosophy entry on fallacies

Drills name only real modes by their UI labels: Case Generator, Brainstorm, Factoids, Flowchart, Case Attacker, Ask Coach, Crossfire and Sparring. They also say where Sparring doesn't run a format.

The shared default FAQs are gone. FAQS and FAQS_DATE are deleted, and the generator now refuses a guide that has no FAQs of its own. The FAQPage JSON-LD is built from the same list as the visible FAQs.

New guard test: src/test/seo/guide-content.test.ts. It runs in npm test and reads public/<slug>/index.html, which the build copies to dist/ unchanged. It checks two things:

• Every guide has at least 700 words in <main> outside the related and About blocks.
• No sentence longer than 60 characters appears on more than 3 guides. The check leaves out nav, the breadcrumb, the footer and the related block.

I confirmed it fails on #233's pages, where it caught 11 thin pages and the shared FAQ sentences on 15 pages, and passes here.

One decision for Adam. Three pages are exempt from the 700-word floor:

• /debatelab-pricing (558 words)
• /debatelab-vs-debateclub (529)
• /debatelab-vs-windebate (654)

The spec didn't scope them, and padding them would mean adding pricing or competitor claims we can't source. The exemption is one Set in the test, so it's easy to remove once those pages get real content.

Word counts

Same method as #233: visible words in <main>, leaving out the breadcrumb, related links and About aside.

Page	Before	After
/debate-clash-points	248	1,222
/how-to-prepare-for-a-debate-tournament	230	1,259
/debate-speech-structure	246	1,215
/debate-fallacies	235	1,099
/debate-motion-analysis	245	1,040
/debate-judge-adaptation	231	1,040
/ai-debate-practice	1,458	1,695
/how-to-practice-rebuttals	1,731	2,033

With #233, all 10 pages in the spec now have more than 1,000 words. Of the 31 remaining guides, only the 3 exempt pages are under 700.

301s, curled locally

I built and served this branch with vercel dev, linked to the staging project, on 127.0.0.1, so vercel.json redirects ran as they do on Vercel:

/debate-collapsing-strategy -> 301 /debate-clash-points
/debate-prep-time-management -> 301 /how-to-prepare-for-a-debate-tournament
/debate-block-file-guide -> 301 /how-to-practice-rebuttals
/debate-pronunciation-and-delivery -> 301 /debate-speech-structure
/ai-vs-traditional-debate-prep -> 301 /ai-debate-practice

All 5 survivors returned 200, and an existing rule (/debate-weighing-guide) still returned 301. Under vercel dev the bare survivor paths served the SPA shell, because Vite handles them. The generated guide HTML is what dist/ serves in production, and check-site resolves every internal link against it.

Sitemap

The build lists 40 URLs, down from 45, so 5 fewer. check-site covers 43 HTML files with no problems.

For Spec 1 (seo/internal-links-brand-entity)

The cluster list in Spec 1 already leaves out all 5 retired slugs, so no cluster changes are needed. Two things to watch:

• Every survivor stays in its cluster: clash points, tournament prep and speech structure in Skills; ai-debate-practice and how-to-practice-rebuttals in Practice and Skills.
• The sections for all 10 rewritten slugs now live in scripts/seo_guides.py, and GUIDES replaces sections wholesale. Add in-body links there, not in PAGES, or they will be silently dropped.

internal_links() is unchanged in both PRs.

Review follow-up

• Deepen the four format guides and shrink the shared About block #233's Policy fix ("2AC answer:") is brought in with a normal merge commit, not a rebase.
• scripts/__pycache__/seo_guides.cpython-312.pyc was committed here by mistake. It's deleted, and .gitignore now ignores __pycache__/ and *.pyc.

Verification

npm run verify ran locally at this head (c349293) and passed. GitHub Actions is billing-blocked, so this is a local pass, not a CI pass.

• docs:check, lint and typecheck passed.
• vitest: 654 passed, including the new guide test. Collaboration tests: 30 passed. Edge tests: 482 passed.
• The build passed. The sitemap has 40 URLs, and check-site found no problems across 43 HTML files.
• test:sw passed, and the Playwright smoke run passed 47 of 47.

In a real browser I checked the stacked tables at 390px and the BP table at 1280px, and confirmed that /tools/debate-timer?format=bp opens with British Parliamentary selected. I didn't run the agentic e2e suite or the verify-debatelab recipes: this PR changes no app flow, and the tool-page edits only change two link targets.

🤖 Generated with Claude Code

[coderabbitai]

Caution

Review failed

The pull request is closed.

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Note

🎁 Summarized by CodeRabbit Free

Your organization is on the Free plan. CodeRabbit will generate a high-level summary and a walkthrough for each pull request. For a comprehensive line-by-line review, please upgrade your subscription to CodeRabbit Essentials by visiting https://app.coderabbit.ai/settings/billing.

_{Comment @coderabbitai help to get the list of available commands.}

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.