Iteration 33 ship log

2026-05-10 · continued autonomous /loop

Date: 2026-05-10 (continued autonomous /loop)

What shipped

9 items. Major link-audit cleanup (95 broken to 15, 84% reduction) + SESSION-MASTER v4.

Link audit deep-dive + fixes

Started iteration at 95 broken internal links. Ended at 15. 80 broken links fixed.

#### Bug 1: foreman-feed slug normalization (20 links recovered)

/srv/sites/factory/foreman-feed/index.html had links like /factory/builds/contract-ai/ (the URL-encoded full URL rendered as a slug). Source: regen-foreman-feed.py extracts slug from foreman report filenames like wishdeal_com_factory_builds_contract-ai-20260510.md but didn't normalize the prefix.

Fixed by adding _normalize_slug() helper in the generator that strips wishdeal_com_factory_builds_ and wishdeal_com_factory_ prefixes. Re-ran. Now links resolve to clean slugs like /factory/builds/contract-ai/.

#### Bug 2: /own/ pages link to non-existent /about/ sub-pages (59 pages affected)

/own/<slug>/index.html rendered two asset cards (Founder persona + buyer-skeptic memos) linking to /factory/builds/<slug>/about/. But promote-artifacts.py only generates /about/ when founder.json exists. Products with skeptic memos but no founder data still got the link card.

Fixed at source in render-own-page.py:

• Skeptic-memos card now links to /skeptic-memos/ (which always exists when the card is rendered)
• Founder persona card now conditional on /about/index.html actually existing

Fixed in-place for the 59 already-rendered /own/ pages with broken links. Quick regex sweep that:

• Replaces the broken /about/ link in skeptic-memos cards with /skeptic-memos/
• Removes the founder persona card entirely when /about/ doesn't exist

#### Bug 3: skeptics page links to /factory/afterhours/skeptic-memos/ (1 link)

/factory/skeptics/ (the skeptics feed) had a "Read full memo" link to /factory/afterhours/skeptic-memos/ which doesn't exist - afterhours skeptic memos live at /factory/builds/afterhours/skeptic-memos/. Source: regen-skeptics-feed.py line 29 special-cased afterhours's product_url to /factory/afterhours/ but the memo URL was built from product_url instead of /builds/<slug>/.

Fixed at source by introducing a separate memo_base_url variable that always uses /builds/<slug>/. afterhours product link still goes to the special promoted page; memo links resolve correctly.

SESSION-MASTER v4

Replaced v3 (which covered iters 1-29) with v4 covering iters 1-32. ~24 hours of session work. Adds entries for iter 30 (4 new audience pages), iter 31 (/factory/adoptability/), iter 32 (/factory/graduated/). 0 em-dashes after sanitization.

Em-dash discipline

0 em-dashes shipped on any iter-33 file. Verified post-each-write.

Files changed inventory

Modified (durable, source-level)

• /home/ubuntu/factory/director/regen-foreman-feed.py (_normalize_slug helper)
• /home/ubuntu/factory/director/render-own-page.py (skeptic-memos card link + founder card conditional)
• /home/ubuntu/factory/director/regen-skeptics-feed.py (memo_base_url separated from product_url)

Modified (in-place patch for already-rendered pages)

• 59 /srv/sites/factory/own/<slug>/index.html (broken /about/ links removed/redirected)

Re-rendered

• /srv/sites/factory/foreman-feed/index.html (cleaner slugs)
• /srv/sites/factory/skeptics/index.html (correct memo URLs)
• /srv/sites/factory/log/SESSION-MASTER.md (v4)

All .bak backups preserved.

Link audit state

Iteration	Broken count	Notes
Iter 33 start	95	baseline
After foreman-feed fix	75	-20
After /own/ /about/ fix	16	-59
After skeptics afterhours fix	15	-1
Final	15	84% reduction

Remaining 15 breakdown:

• 6 JavaScript template literals (${slug}, ${p.slug}): false positives, work at runtime
• 2 stale references (outbound-cadence-ai, sc-insights): products that were removed/renamed; the references are in admin/log pages, not visitor-facing
• 5 missing product sub-pages (afterhours/integrations, afterhours/security, handoff-ai/pricing, photo-ai/pricing, etc): products that don't have those particular sub-pages
• 2 missing /sales-kit/ (customer-data-platform, catering-ai)

All remaining are in admin/log pages or are false-positive JS template literals. None are in primary buyer-facing surface.

Why this matters

A buyer browsing 192 ideas should never hit a dead-end click. Iter 26 fixed the missing top-nav class of issue (1500 pages). Iter 33 fixed the "links that exist but point to nothing" class (60+ links touched). Both are quality signals - a buyer looking around for a few minutes should not find a single broken link in primary surfaces.

Status snapshot

• 192 products in catalog
• 62 buyer-facing pages
• 10 audience landing pages
• 59/59 health endpoints passing, avg 3ms
• Sitemap: 1328 URLs
• 0 em-dashes shipped across 33 iterations
• Link audit: 15 broken (all in admin/log/JS-templates, none in primary buyer surface)

What still needs work

Same Wes-side as before:

1. Stripe wiring (30 min)
2. Email-send for auto-fulfill
3. First real traffic push

Next 5 recommended autonomous tasks

1. Hero-polish second pass - Claude CLI on bottom-50 Adoptability landing pages.
2. Suppress link-audit false positives - update link-audit.py to skip URLs inside <script> tags (the JS template literals are noise).
3. Operator inbox enhancement - per-product CTR calculation when intent captures arrive.
4. Surfacing the trust signal block on /unlock/ pages too - currently only on /builds/, but unlock decision could also benefit.
5. Per-product /how-it-works/ page - lightweight new content surface; pulls implementation_plan + tech_stack.

Cumulative iter 1-33

The factory's buyer-facing surface is now genuinely tight. 95 broken links discovered, 80 fixed (84% reduction), no em-dashes shipped, ~1500 sub-pages with consistent nav, 191 of 192 products with all 11 injection markers, 59 of 59 health endpoints green, 10 audience entry points, per-product previews + objection handlers + trust signals + graduated proof page. Remaining work is real customer feedback (requires Wes-side traffic push or Stripe).