Iteration 71 ship log

2026-05-14 · push mode, 30-min cadence, NEW VALUE pivot

Date: 2026-05-14 (push mode, 30-min cadence, NEW VALUE pivot)

What shipped (3 substantive ships)

The audit-arc was clean enough by iter 70 that iter 71 pivoted to new-value mode. Three ships:

1. NEW playbook essay: "How we caught 70 fabrications in our own catalog" (1800 words)
2. NEW Stripe wiring technical notes doc for when Wes is back
3. Playbooks index intro refresh to reflect 9 essays / ~16,500 words

Ship 1: NEW essay - "How we caught 70 fabrications in our own catalog"

Live at https://wishdeal.com/factory/playbooks/seventy-fabrications/. The 9th playbook essay. Different shape from the previous 8: this one is an admit-and-explain piece about the studio's own failures.

Structure:

• "What we shipped that was not true" - full inventory in a table, by iter (65 to 70), showing how many fabrications were caught in each, where they lived (FAQ subpages, pricing pages, audit-ai placeholders, enterprise template, service-business pages), and what their source bug was
• "Why the bugs survived the prompt" - 4 root causes: (1) LLM creative-route around the prompt, (2) different generators with different prompts, (3) "skip if file exists" preservation logic preserving deprecated content, (4) hardcoded template-level fabrications
• "How we found them" - the fix-and-find loop pattern
• "What this costs us to admit" - operator-honest framing about hurting to publish
• "The general lesson for any autonomous content system" - 4 practices for anyone running an LLM content pipeline
• "What to look for next" - categories of bugs not yet audited

Why this essay matters: The Wishdeal Factory marketed itself as honest from day one. The /factory/honest/ page says "the median idea has zero customers, zero validated demand signal." That's an aspirational claim until you prove it under audit pressure. This essay is the proof. It says specifically: we shipped 70 fabrications, here's the inventory, here's what we did about it. That kind of transparency is the credibility surface that competitors won't write.

Wired into:

• /factory/playbooks/index.html (new card inserted above the iter 69 "skip-these-dossiers" essay)
• regen-sitemap-v2.py (+1 URL, sitemap now 2313 URLs)
• regen-health-check.py (+1 endpoint, now 70/70 passing)

The essay positions Wishdeal Factory as the rare autonomous-content shop that publishes its own audit results.

Ship 2: Stripe wiring technical notes (Wes-blocker prep)

Live at https://wishdeal.com/factory/log/STRIPE-WIRING-NOTES.md.

A focused technical brief for what Wes needs to do when he's back to enable the $5 unlock flow. Currently catalog says "Coming soon" because FACTORY_STRIPE_UNLOCK_URL env var is unset.

Content covers:

• What exists already (unlock pages, generator gating, HMAC tokens, no webhook handler)
• 6 steps in order with time estimates: Stripe account setup (10 min), Payment Link creation (5 min), env var (1 min), webhook handler (15-20 min), success page (5 min), test->live cutover
• Path A (Payment Links, no code) vs Path B (server-side Checkout Sessions) recommendation with reasoning
• Expected blockers + workarounds (Stripe CLI for local testing, SMTP for email deliverability, token mint retry logic)
• Total: ~45 min for Path A v1

The goal: reduce Wes's "back from away" to "first $5 unlock" time to under 1 hour. The doc names every env var, every endpoint, every Stripe Dashboard step.

Ship 3: /factory/playbooks/ intro refreshed

Now reads: "Nine essays, ~16,500 words" (was "Seven essays, ~13,000 words"). Reflects the iter 69 "skip-these-dossiers" + iter 71 "seventy-fabrications" additions. One-line surgical fix via sed.

Files changed inventory

New (durable)

• /srv/sites/factory/playbooks/seventy-fabrications/index.html (new essay)
• /srv/sites/factory/log/STRIPE-WIRING-NOTES.md (Wes-blocker doc)

Modified (source-level + in-place)

• /srv/sites/factory/playbooks/index.html (new card + count refresh)
• /home/ubuntu/factory/director/regen-sitemap-v2.py (+essay URL)
• /home/ubuntu/factory/director/health-check.py (+essay endpoint)

Status snapshot

• 238 products, 0 broken pages
• ~0 fabrications catalog-wide (70+ cleared across iters 67-70, durable cron in place)
• 9 substantive playbook essays totaling ~16,500 words
• 5 foundational high-trust pages depth-passed (complete)
• 60 bulk-repaired + 4 hand-repaired + 9 polished + 2 confirmed-good + 1 audit-fix
• 10 content invariants defended at surface+source
• audit-fakeproof.py runs daily at 4:30am (durable safeguard)
• 2313 sitemap URLs
• 70/70 health endpoints passing (added new essay)
• 0 em-dashes shipped this iter

The 9-essay playbook library

#	Iter	Essay	Words	Shape
1	46	$5k budget	1800	Tactical
2	47	Vertical AI 2026	1800	Market thesis
3	48	Operator-partnership math	1850	Tier math
4	49	Agency productization	1900	Segment
5	50	Distribution channels 2026	1950	Distribution
6	51	Counsel anatomy	1700	Case study
7	52	Five patterns	2050	Editorial
8	69	Skip these dossiers	1700	Anti-recommendation
9	71	Seventy fabrications	1800	Studio honesty

The library now covers tactical / strategic / methodological / editorial / honest-disclosure / studio-transparency angles. A reader who reads all 9 understands the catalog from every operator perspective.

Running queue (top 5 for iter 72)

1. Open Graph image generator for the 9 essays + 5 foundational pages (deferred from iter 70/71, social shares look bare without OG images)
2. /factory/builds/audit-ai/ screenshot repair (cron failure, low priority)
3. Per-product polish on remaining 70-tier products (campaign-budget-ai 71, supplier-ai 71 already polished, lead-router 71 clean, but others remain)
4. Investigate the 5 remaining audit-fakeproof.py hard findings (likely all false positives now)
5. /factory/cron-status/ page for transparency about which crons are running

Cumulative iter 1-71

• Catalog: 238 products, 0 broken, ~0 fabrications
• Content library: 9 essays / ~16,500 words covering 9 distinct angles
• Foundational pages: 5 of 5 complete and depth-passed
• Hand-polished products: 9
• Source durability: 14+ generators source-fixed + audit-fakeproof.py in cron
• Content invariants: 10 defended at surface+source levels
• Wes-blocker prep: Stripe wiring notes ready for ~45-min path-A v1 implementation

The factory is structurally complete and content-honest. The work that remains is per-product polish + Wes-side blockers (Stripe, email, traffic).