Rot vectors and where to look first

The system is single-operator and accreted across many decisions. Some coupling is deliberate (storage routing), some is the cost of not having a v2 refactor yet. This page documents both — what to watch for, and where to start when something changes.

Adding a new domain — the 6 touch points ! The 6-touch-point rule is conservative. Some changes cascade beyond these six. A storage backend change for an existing door (e.g. moving learning to a DB) would also require migration schemas, a new reconciler branch, and possibly an RLS policy.

The model implies "just add an entry to _domains.yaml." Reality is six places, of which only one is declarative:

Hardcoded coupling — find before refactoring

Renaming a domain (e.g. finance → markets) requires touching every place the literal string appears. Inventory of known sites:

Known limitations to plan around

What the deep-retrieval & de-biasing program moved

The 2026-06-13 retrieval-depth program retired or mitigated several of the rot vectors this page used to list as open:

New rot vectors the program introduced

Every fix has a cost. These are the new coupling/fault-lines to watch:

• Cross-repo payload contract. Deep results land in three stores under one shape — TradingV Postgres (/v1/rx/deep), Supabase, and a learning markdown sidecar — plus the separate fitness-UI repo. A schema change must move all of them together.
• Two identity spaces. The Supabase owner id (verocity auth uid) and the TradingV operator id are different values. A wrong-id write is invisible to RLS-scoped reads.

Landmine ledger — deferred debt with explicit triggers

Tracked in the rx repo's tech-debt log; each entry is fine at n=1 and has a named condition that should force the rework:

Where the doors-storage mapping actually lives ! The mapping {fitness: supabase, nutrition: supabase, finance: tradingv_local, learning: markdown} is not declared in any single yaml or json file. It is reconstructable only by reading D-045 + D-047 in rx-meta/DECISIONS-LOG.md plus the trading app's internal project docs. A maintainer wishing to verify "where does door X write?" must read those prose sources, then grep slash commands.

The closest thing to a canonical declaration is the facts inventory in the project README.md (alongside this site) and the source-of-truth table in AGENTS.md §3. If you change the mapping, update those tables first — they are the only spots designed to be re-quoted.

How this site stays honest — the re-verification workflow

This explainer is documentation about other systems, so it rots the moment a source repo changes. Rather than trusting the pages, every update runs a re-verification pass. First, each upstream surface is independently re-researched: the trading app (tables, ingest sources, endpoints), the four door workspaces (rx artifacts and the decisions ledger), the knowledge-vault (_domains.yaml, embedding pipeline, taxonomy), and the Supabase schema behind the verocity app.

The findings then face an adversarial review from four deliberately opposed perspectives: a skeptic hunting factual errors, an architect hunting missing edges and failure modes, a user advocate checking clarity for non-technical readers, and a maintainer checking rot vectors and hardcoded coupling. Only pages whose facts actually changed get edited, every load-bearing fact keeps a pointer to its canonical source, and the source repos themselves are never touched from here — this site is additive only.

Source-of-truth quick reference

The deletions log — what got removed

Architectural simplifications worth remembering:

Future v1.x candidates (per ROADMAP-v1_x) ! What is intentionally not in scope. Multi-operator support, fancier RLS, real-time cross-domain orchestration, scheduled auto-rec firing without command invocation, a v2 of the embedding model. Each of these has a v1.x or v2 placeholder but is deferred deliberately until the action-rate signal proves the system is worth investing in further.

• /rx-meta — cross-domain meta-recommender (Phase O). Unlocks on ≥30% action-rate on every gated door plus ≥10 dispositioned recs total across domains (D-030).
• Phase K scheduler — auto-trigger recs on threshold breaches without manual command invocation. Gated on action-rate health.
• Explicit hypothesis ↔ rec frontmatter linking (replaces substring heuristic).
• Nutrition rx generator wiring (Phase J.2) — depends on operator-chosen daily-log source.
• Learning corpus ingest at scale — depends on operator drop of URLs.