Scaffold ~/rdco-mac/ Layer 1 playbook repo first (the thing being sold doesn't exist as code yet),
then ship sc.raydata.co/mac LP + Day 1 drip live (the productization surface that unblocks
reader sensor + funnel simultaneously). Stripe→GITHUB_TOKEN flow follows once LP captures
first interested-buyer signal. Severity-tier actuator and matrix-fill copilot are downstream
of having any paid users to validate against.
Progress engagement produces lived diagnostic per sprint (what hurts, what surfaces real
bugs, where matrix is overkill) but diagnostic isn't structured back into framework
refinement on a cadence. Case-study artifact exists for gold_opp_pipeline; no ongoing
loop says "this week's Progress findings → next testing-matrix-template revision →
updated /audit-model skill prompt."
The productized loop (reader signal → article/drip refinement → conversion) doesn't
exist at all because the public surface doesn't exist.
10 gaps — show
highAdoption signal from external dbt teams
No instrumentation on whether anyone outside founder uses the matrix. No download counter, no GitHub stars, no inbound questions tagged MAC.
mediumTime-to-fill-matrix per model
Matrix takes hours to fill. No measurement of how long, which cells get skipped,
or which cells produce highest-value findings. Friction sensor is qualitative-only.
highFalse-alarm rate on severity tiers
Stop/Pause/Go severity is the load-bearing claim. No measurement today of how often
Stop alerts are real vs noise on Progress. If tier mapping is wrong, the entire
"agents triage, humans review exceptions" promise breaks silently.
highDrip-course / anchor-article reader signal
Drafts exist (mac-anchor-article-draft-v1, drip-day1) but no published surface,
no LP, no email capture. Zero sensor on whether productized path resonates.
Inherits cleanly from SC's email-engagement-instrumentation when that ships.
highMAC landing page / public surface (sc.raydata.co/mac)
CRITICAL COMPONENT. No public surface yet. Blocks BOTH the productized funnel
AND the reader-signal sensor. Modular reuse target: /build-landing-page skill +
ray-data-co-design + sanity-check-design umbrella. Inherits SC's site infra.
highDrip-course delivery infrastructure (Resend wiring for MAC)
Resend wiring exists for SC; no MAC-specific drip sequence provisioned.
Modular reuse — parameterization of SC's mailing-list capability, not net-new.
Day 2-7 drafts still need writing (Day 1 drafted 2026-04-15).
DECIDED 2026-05-04 distribution shape: buyer pays via Stripe payment link →
webhook generates scoped GITHUB_TOKEN → token + install instructions emailed →
buyer runs `claude /plugin install raydataco/mac`. Updates ship via git push.
Anthropic takes no cut (out-of-band monetization).
highLayer 1 playbook scaffold (~/rdco-mac/ — the portable IP)
The actual 8-phase agent playbook. Portable across runtimes (Layer 1) with
Claude skill wrapper (Layer 2) as v1 distribution. This is the thing being sold.
Framework specifies Stop/Pause/Go but no skill/tool takes dbt run output, classifies
failures by tier, and routes accordingly. "Agents triage" claim is aspirational, not
implemented. Coupled with false-alarm-rate sensor — sensor without actuator is wasted.
mediumMatrix-fill copilot (pre-populate cells from schema YAML)
Filling matrix is the friction. No tool that pre-populates likely cells from
a model's schema YAML, leaving engineer to confirm/edit rather than start blank.
Unlocks adoption velocity for both consulting and productized paths.
Executable info-product shape DECIDED — $300-2k Claude plugin via Stripe paywall, two-layer portable IP + skill wrapper architecture
Retainer/consulting framing killed via HQ click-back rail (first real round-trip on /decisions surface). phData conflict-of-interest was load-bearing reason.
Four-layer audit filed; three load-bearing gaps named (productization surface, severity-tier actuator+sensor, matrix-fill copilot)