# Product Operating System — Pulling It Together (v1)

**Date:** 2026-09-26  
**Status:** Active operating plan (living)  
**Owner:** James (decisions) · Agents (drafting / execution assist)  
**Live pages:**
- [HTML hub (gt_dev)](https://smb.a1prospect.com/docs/PRODUCT_OS.html)
- [HTML hub (system)](https://trsportsacademy.com/system/PRODUCT-OS.html)
- [This markdown](https://smb.a1prospect.com/docs/PRODUCT_OS_v1.md)

**Canonical siblings (do not replace — compose):**
- [UNIFIED-FRAMEWORK-v1.md](https://trsportsacademy.com/system/UNIFIED-FRAMEWORK-v1.md) — doctrine (spine / cockpit / lens)
- [AMS-INTEGRATION-PLAN-v1.md](https://trsportsacademy.com/system/AMS-INTEGRATION-PLAN-v1.md) — AMS glue & phases
- [AMS-COMMERCIAL-ROADMAP.md](https://trsportsacademy.com/system/AMS-COMMERCIAL-ROADMAP.md) — SaaS / entitlements vision
- [System hub](https://trsportsacademy.com/system/) — 13-property production index
- Adaptive Mean rebuild: `C:\Users\jtdom\adaptive-mean-sigma-bands\docs\PLAN.md`

---

## 0 — Verdict

**Yes — this makes sense.** You already have most of the pieces as scattered lenses and cockpits. What is out of control is not capability; it is **lack of a single product spine, hard sequencing gates, and WIP discipline**.

The product end-state is:

> **AMS is the dashboard shell.** Each proven URL becomes a **service offering** behind entitlements. Strategies are first-class objects (shareable, gradeable). The live day loop is: plan → check go/no-go → monitor setup → manage risk vs daily stop → review.

This document is the **operating plan** that sequences work so we stop thrashing across indicators, Lens, Scorecards, AMS, SIP, MGP, Multi-TF, and reports at once.

---

## 1 — North star (mapped to product capabilities)

| Your goal | Product capability | Primary surfaces today |
|---|---|---|
| Proven strategies + user-owned strategies | Strategy Library + ACL/share | `/strategies/`, `strategies.json`, playbooks |
| Share strategies with access control | Entitlements + strategy ACL | AMS `ams.services` / future strategy grants |
| Grade trade vs strategy (go / no-go) | Pre-trade grader / Lens grade | Lens (`playbooks/lens.html`), scorecards, SIP |
| Monitor trades during the day | Live Lens + HUD | live-lens-preview, accountability HUD |
| Running P&L + status to daily stop | AMS / Accountability cockpit | `/ams/`, live-accountability-v2/hud, daily-stop |
| Open risk vs daily-stop breach | Risk-at-open vs stop remaining | HUD + daily-stop + open positions |
| Market / sentiment through the day | Environment / Multi-TF stack | multi-tf-stack, breadth, MGP, live classifier |
| Few high-signal alerts (or owned queues) | Alert policy + owners | webhooks/v2-gpt/rt, indicator alerts, AMS alerts |

**One-sentence product thesis:**  
Indicators and reports are **primitives**; strategies are **contracts**; AMS is the **product shell**; everything else is a **service card**.

---

## 2 — Inventory (known surfaces → service slots)

Hosts: **smb** = `smb.a1prospect.com` → `~/gt_dev/` · **trsa** = `trsportsacademy.com` → `~/trsports_gd/html/` · **dev.gt** = `dev.grumpytrade.com`.

| Service slot | Role | Example URLs / paths | Docroot |
|---|---|---|---|
| **AMS shell** | Product launcher / auth / entitlements | `/ams/`, `/ams/system/`, `/ams/app/` | gt_dev → trading/dev/ams |
| **Accountability HUD** | Cockpit: P&L, rules, day status | `/live-accountability-v2/hud/` | gt_dev |
| **Daily Stop** | Cockpit: stop limit | `/daily-stop/` | public_html |
| **Strategy Library** | Doctrine registry | `/strategies/` | public_html |
| **Playbooks / Lens** | Teach + grade + replay | `/playbooks/lens.html`, `/playbooks/daily/` | both |
| **Scorecards** | Post-trade execution grade | `/scorecards/` | both (symlink) |
| **MGP** | Morning plan | `/mgp/`, trsa `/mgp` | both |
| **SIP** | Setup / interest plan pages | `/sip/` | public_html |
| **MA Flat / AM Sigma report** | Indicator → report lens | `/maflat/`, `report-am-sigma.html` | gt_dev / dev.gt |
| **Live Lens preview** | Intraday setup monitor | `/live-lens-preview/` | gt_dev |
| **Multi-TF stack** | Market structure / sentiment | `/system/multi-tf-stack/` | public_html/system |
| **Guides** | Trader docs | `/docs/AM_Sigma_Trader_Guide.html`, VWAP Shannon Fade | gt_dev/docs |
| **DB reports** | Research lenses | `/db_reports/...`, fashionably-late, vwap-bands | gt_dev |
| **GapMA / PBS / PVP** | Specialized scanners | `/gapma/`, `/pbs/`, `/pvp/` | gt_dev |
| **Webhooks / RT** | Alert ingress | `/php/webhooks/v2-gpt/rt/` | gt_dev |
| **Calendar / audit / MGP (trsa)** | Ops / review | trsa `/audit/`, `/mgp`, jd/cal-next | mixed |
| **Indicators (source)** | Chart truth engines | `Z:\trading\indicators`, local `adaptive-mean-sigma-bands` | code |

**Assumption:** more surfaces exist (backups, `*-dev`, retired trees). Phase A freezes the **canonical** list; everything else is tagged `archive` / `dev` / `retire`.

---

## 3 — Architecture reminder (do not reinvent)

From Unified Framework — keep this mental model:

| Layer | Meaning | Failure mode |
|---|---|---|
| **Spine** | PostgreSQL facts (trades, alerts, users, settings) | Everything dies |
| **Cockpit** | Daily driver UI (HUD, daily stop, risk) | Trade blind for a day |
| **Lens** | Analytical views (scorecards, playbooks, reports, MGP) | Rebuildable in a weekend |

**Trader lifecycle (left → right):**  
Strategy Library → Playbooks → MGP → Accountability HUD → Scorecards → Monthly → Param Surface

AMS wraps this lifecycle as **entitled services**, not as a rewrite of each lens.

---

## 4 — Sequencing: hard gates vs parallel tracks

### Rule

- A **hard gate** must be green before the next phase starts **as a team focus**.
- **Parallel tracks** may run inside a phase if they do not mutate the same source of truth.
- **WIP limit:** max **2 active focus tracks** + **1 bugfix lane**. Everything else is parked with an owner and a next review date.

### Phase map

```
A Inventory freeze ──► B Truth engines ──► C Strategy contract ──► D Live day loop
                              │                      │                     │
                              ├─ Indicator (AM σ)    ├─ Grader API         ├─ Lens live
                              ├─ Identity / PnL      ├─ Share ACL          ├─ Risk-at-open
                              └─ Alert schema        └─ Proven list        └─ MGP ↔ HUD
                                                                            │
                                                                            ▼
                                                                   E AMS product shell
                                                                            │
                                                                            ▼
                                                                   F Alerts & sentiment
```

### Phase A — Inventory freeze & ownership *(gate: catalog signed)*

**Must finish before moving on as a team:**

1. Canonical service catalog (this table, completed) with **one owner** per service.
2. Tag every known URL: `canon` | `dev` | `archive` | `retire`.
3. Declare **source of truth** per concept (PnL, daily stop, strategy id, alert id).
4. Working agreements locked (see §6).

**Parallel OK:** continuing bugfixes on a single owned surface; no new “side projects” without catalog row.

**Exit criteria:** signed catalog + owners; no orphan URLs in the “we use this daily” set.

---

### Phase B — Truth engines *(gate: chart + ledger trustworthy)*

**Hard dependencies (serial within B where noted):**

| Order | Workstream | Why first | Exit |
|---|---|---|---|
| B1 | **Adaptive Mean + σ Bands** engine correctness (triggers, voids, TRADE vs INFO, labels) | Everything that grades/monitors setups inherits chart lies | Live chart matches intent; known issues closed or deferred with IDs |
| B2 | **Identity function** `account_id → user_id → trader_name` used everywhere | Without it, AMS scope and scorecards disagree | One helper; consumers listed |
| B3 | **PnL / open risk spine** (HUD + daily stop + open stops) | Accountability product promise | Running P&L and risk-to-stop agree with ledger |

**Parallel inside B (after B1 started):**

- Docs / tooltips for AM σ (does not change engine math).
- MA Flat / report generators **read-only** against frozen indicator version.
- Alert DDL / freshness dashboard (read paths).

**Do not start yet:** new strategy ports to KITE, Shannon-drift merges into AM baseline, new alert spam channels.

---

### Phase C — Strategy contract *(gate: gradeable go / no-go)*

**Depends on:** B1 (setup definitions) + strategy schema in `/system/`.

| Workstream | Parallel? | Deliverable |
|---|---|---|
| Proven strategy list (curated subset of registry) | with schema polish | “Ship these first” list |
| Strategy schema fields needed for grading | serial with grader | Entry / stop / target / invalidation / env gates |
| Pre-trade grader (go / no-go + insight) | after schema freeze | API + Lens/UI call site |
| User strategies + share ACL | parallel with grader **UI**, after schema | Create / share / revoke |

**Exit criteria:** for ≥1 proven strategy, a trade (or proposed entry) returns a deterministic grade with reasons.

---

### Phase D — Live day loop *(gate: one coherent session)*

**Depends on:** B3 (risk) + C grader (for go/no-go) + MGP/SIP as inputs.

| Slot | Surface | Job in the loop |
|---|---|---|
| Pre-market | MGP + SIP + Multi-TF | What matters today |
| Entry | Lens grader + strategy ACL | Go / no-go |
| Intraday | Live Lens + Multi-TF | Setup still valid? |
| Risk | HUD + daily stop + open risk | Can I hold / add / must flatten? |
| Post | Scorecards | Did I follow the contract? |

**Parallel OK:** Multi-TF polish, MGP tone, SIP page UX — as long as they **consume** shared env tags, not invent competing regimes.

**Exit criteria:** one trader can run a full session using only AMS launcher → linked services without hunting URLs.

---

### Phase E — AMS product shell *(gate: services as offerings)*

**Depends on:** Phase D loop proven for internal users; existing AMS auth/entitlements.

| Workstream | Notes |
|---|---|
| Two-zone launcher (active vs catalog) | Per commercial roadmap |
| Each catalog row = service card → deep link | Wrap legacy in shell; do not rewrite |
| Entitlements per service | Already seeded; wire remaining surfaces |
| Strategy share uses same ACL patterns | Do not invent a second permission system |

**Parallel OK:** marketing copy / detail pages; DKIM; invite flows.

**Exit criteria:** every `canon` service is reachable from `/ams/` with entitlement check (even if legacy UI still loads underneath).

---

### Phase F — Alerts & sentiment ops *(gate: signal, not noise)*

**Depends on:** Phase D (otherwise alerts have nowhere useful to land).

| Workstream | Rule |
|---|---|
| Alert taxonomy (setup / risk / market / system) | Max N alerts per type per hour |
| Owner per queue (or mute) | Unowned alerts = off |
| Multi-TF / breadth as market sentiment service | Feeds MGP + HUD badge, not a third PnL |

**Exit criteria:** alert policy doc live; default user sees ≤ a small curated set.

---

## 5 — Parallelism matrix (quick reference)

| Track | Can run with | Blocked by |
|---|---|---|
| AM σ indicator correctness | Docs, MA Flat read-only reports | — (start now) |
| Lens UX / replay | Playbook content | Grader contract (for go/no-go claims) |
| Scorecards | Report polish | Identity fn + PnL spine |
| AMS shell / entitlements | Landing, admin UI | Full wrap of cockpits until B3 stable |
| MGP / SIP content | Multi-TF readouts | Competing regime definitions |
| Multi-TF stack | Breadth | Becoming a second accountability UI |
| KITE / backtest ports | Research only | **Hard gate:** AM σ live truth (B1) |
| New scanners (gapma/pbs/pvp) | Parked unless catalogued | Phase A ownership |

---

## 6 — Working agreements (how we execute)

| Concern | Rule |
|---|---|
| **a2 vs Z:** | `ssh a2` for search/read/write on the share. `Z:\` = same files via SSHFS — slow; avoid recursive scans on Z. |
| **Paths** | On a2: `/home/trsports/...` (= `~/...`). There is **no** `/z` on a2. |
| **kti / Einstein / Windows apps** | Execute on a Windows workstation; read/write artifacts via a2 when files live on the share. |
| **gt_dev vs public_html** | smb → `~/gt_dev/` · trsa → `~/trsports_gd/html/` (symlink `public_html`). Prefer additive changes; AMS parallel-build rules still apply. |
| **Atlas / Claude memory** | Treat Atlas + `~/.claude` memory + this Product OS as the map; do not start a parallel “secret” plan in a scratch folder. |
| **HUD edits** | Live HUD remains coordination-sensitive; prefer `*_dev` / AMS wrap until promotion. |
| **Charts** | Shared TradingView widget or `/_shared/setup_chart.js` — no new chart forks. |

---

## 7 — Project management cadence

### WIP and board

Simple board columns (use `/todos/` or a PRODUCT_OS status table on the HTML page):

`Backlog` → `Ready` → `Doing (max 2)` → `Blocked` → `Review` → `Done (this week)`

Each card needs: **owner**, **phase (A–F)**, **depends_on**, **definition of done**, **canon URL**.

### Weekly rhythm

| When | Ritual | Output |
|---|---|---|
| Mon | Pick ≤2 focus tracks for the week | Update PRODUCT_OS status table |
| Daily (async) | Blockers only — what stopped a gate | Blocked column |
| Wed | Mid-week gate check | Keep / slip / park decision |
| Fri | Demo one user-visible slice | Link + screenshot in notes |
| Fri | Park list | Anything not in focus gets a next-review date |

### Definition of Done (any card)

1. Lives under a **canon** path (or explicit `dev` with promote plan).  
2. Has an owner and is linked from AMS catalog **or** Product OS inventory.  
3. Declares data source (spine table / API).  
4. Does not invent a second PnL, identity, or regime without ADR.

---

## 8 — Immediate focus (next 2 weeks)

**Focus track 1 — Phase B1:** Adaptive Mean + σ Bands correctness + label/tooltip clarity (`adaptive-mean-sigma-bands` PLAN.md).  
**Focus track 2 — Phase A:** Finish service catalog tagging + owners for every URL in §2 (and discoveries).  

**Bugfix lane:** only production-breaking issues on HUD / MGP / scorecards.

**Parked until B1 green:** KITE ports of coil/expansion, Shannon-drift merges into baseline, new scanner products.

**Parked until B3 green:** promising “risk at open vs daily stop” as a marketed AMS feature (design OK; ship after spine agrees).

---

## 9 — Success metrics (product, not vanity)

| Metric | Target |
|---|---|
| Canon services linked from AMS | 100% of daily-driver set |
| Orphan “we use this” URLs | 0 |
| Competing PnL numbers across HUD / scorecard / reports | 0 |
| Strategies with automated go/no-go grade | ≥1 proven, then expand |
| Active focus tracks | ≤2 |
| Unowned alert types in default view | 0 |

---

## 10 — Change log

| Date | Change |
|---|---|
| 2026-09-26 | v1 created — inventory, phases A–F, WIP rules, working agreements, 2-week focus |
