STUD.com · the shape of the work

System map: five layers,
one dogfood loop

How the pieces relate. A process layer produces the locked strategy, which constrains the built engine, which serves the curated supply, which renders in the front. And the project builds itself with its own map. Read top to bottom; the verbs on each connector are the relationship.

built / verified gated on creds gap / not yet built the live loop

Process / founder-OS

wraps all how the work gets made

Decision process

draft → 6-lens probe → lock

Curation pipeline

mine → lens → objective → slice

Baton + handoffs

session-to-session continuity

STATUS / CHANGELOG / plans

the living state files

/probe /foresight /baton

the skills (proto-STUD, by hand)

AI operating system

loops · memory · 7 audit realms · the cadence tick

produces

Strategy & decisions

locked the why and the what; the source of truth

v3.0 build spec

the adopted, consolidated v1 architecture

DECISIONS.md

the D-001..D-020 ledger

Positioning compass

map + trusted execution; the wedge

Roadmap

the phased critical path

constrains

The engine

built the trust machinery, verified without creds

D-004 spine

Objective / Principal / Operation + append-only acceptance

Orchestration

start → judge → tiered settle

Judge: machine tier

held-out-case oracle (slugify proven)

Judge: external tier

predicate over a signal (deploy-health proven)

§4.6 isolation

process-isolated grader, attack-fixture-proven (1591 + 44 checks)

Published interface

every contract publishes the enforced interface, drift-gated

Probe adapters

gated: real HTTP / CVE / deploy + creds

serves

Content / supply

wip · active the curated map + the authored objectives

Mining

7 competitor IAs + synthesis

Lenses + playbooks

role / situation / goal (14 live playbooks)

Objectives + contracts

262 settle live (74 machine, 187 doc-validators, 2 external)

Sellable slices

9 slices incl. the dogfood checklist

Founder-venture map

the 6-question blindspot backlog

Content types

objective / playbook / framework

Framework / auto tier

gap: empty by design (mining was settleable-biased)

renders in

The front

live what a visitor sees at stud.com today

Platform UI @ stud.com

the live soft launch: search-first, 262 objectives, paid fulfillment gated

Email capture

api/signup → Neon (live)

Preview SPA

the rollback path (disposable)

Status & gaps

Built + verified

  • D-004 spine & D-002 settlement skeleton
  • Machine + external acceptance tiers + §4.6 isolation
  • 20 locked decisions + the v3.0 spec
  • The search-first front + start flow + 262 objectives settling live
  • The machine interface published on every contract (drift-gated)
  • The AI operating system: loops, memory, 7 audit realms
  • 14 playbooks, the map + type system; Neon + GitHub sign-in live

Gated on creds

  • Stripe Billing live keys (post-LLC)
  • D-002 counsel-gate (before the first live txn)
  • Judge prod wiring: shared secret + transport URL
  • Prod catalog edits via sync-catalog (Dan-gated)

Gap / not yet built

  • Verified-outcome track records (post-launch)
  • Ordered-playbook + framework / auto tier (empty by design)
  • Live dogfood probe wiring (adapters built)
  • Flavor-b run-then-measure (coverage)
  • Runtime sub-commissioning (D-010, deferred)

Highest-value next

  • Open paid fulfillment: Stripe live keys + the counsel gate
  • → the 13-playbook dogfood sweep (inputs green)
  • → the first real transaction; grow the catalog

Source of truth: SYSTEM-MAP.md (annotated tree + flows + processes) · current work: STATUS.md · locks: DECISIONS.md
A living map, refreshed 2026-07-07. Refresh at checkpoints as the layers fill in.