# SFIB Operations · the spaghetti, decomposed **Owner**: Shalaco · **Last edit**: 2026-05-04 · **Repo**: code-compendium-26 A single brief that says: what we've built, what's live, what's next, and in what order. Replaces the implicit roadmap that's been living in chat context. --- ## 1 · The mission Build SFIB an operator-readable control system covering: | Layer | Question it answers | |---|---| | **Pricing** | What price for each (mix × format) — locked, defensible, year-over-year stable? | | **Inventory** | What do I have, what's sold, what's listed, what's incoming, what's overcommitted? | | **Purchasing** | What do I need to buy this week to maintain velocity? | | **Sales** | Who's our top customer / mix / SKU? Where's the volume? | | **Operations** | Which orders are late? Which batches are open? Who's assembling what? | | **Forecasting** | When will I run out at current pace? When will incoming arrive? | The deliverable surface is `wildfag.com/` (password-gated) — a constellation of self-contained HTML dashboards driven by exported CSVs from Squarespace, ShipStation, supplier invoices, and operator inventory snapshots. --- ## 2 · What's live right now ### On wildfag.com (production, password-gated) | Page | What it does | Status | |---|---|---| | `index.html` | Landing page · 9 dashboard tiles | ✓ live | | `monday-morning.html` | Daily operator briefing | ✓ live (SLA threshold is a placeholder, see §4) | | `purchasing.html` | What-to-buy-today, sorted by velocity | ✓ live (NEW today) | | `inventory-dashboard.html` | Editable 16×9 stock grid + CSV export | ✓ live | | `inventory-flow.html` | Time-series flow + balance bars + net position | ✓ live | | `sfib-matrix.html` | 16×9 sales × stock × margin matrix with product images | ✓ live | | `recipe-matrix.html` | BOM reference: SKU + seed oz + sq-ft per cell | ✓ live | | `pricing-table.html` | 80-row flat metrics (retail, COGS, margin, vs-market) | ✓ live | | `pricing-lock-v2.html` | 16×5 toggleable retail/COGS/margin matrix | ✓ live | | `sq-import-audit.html` | Validate SQ products CSV before import | ✓ live | ### Source-of-truth files (in repo, not on web) | Path | Role | |---|---| | `05-02 SFIB SKU Matrix/outputs/cogs-matrix.csv` | 80 shaker SKUs, locked formula prices + COGS | | `05-02 SFIB SKU Matrix/outputs/squarespace-import-shakers.csv` | 144-variant SQ import (SQ8- prefix) | | `05-02 SFIB SKU Matrix/outputs/flow-data.json` | Per-mix timeline (stock/incoming/sold) for inventory-flow page | | `05-02 SFIB SKU Matrix/inputs/cost-data/inventory-2026-04-28.md` | Seed inventory snapshot from xlsx | | `05-02 SFIB SKU Matrix/inputs/cost-data/incoming-shipments.csv` | Supplier POs not yet arrived (NE 2×5lb) | | `05-02 SFIB SKU Matrix/inputs/cost-data/competitor-pricing.md` | American Meadows + Nature's Seed market refs | | `05-02 SFIB SKU Matrix/inputs/cost-data/pcs-invoices.csv` | Supplier purchase log | | `001 Active Scripts/Winventory 2026/variant_key.csv` | Gold-standard SKU → mix/format mapper (175 rows) | | `05-02 SFIB SKU Matrix/scripts/orders-to-flow.py` | Generates flow-data.json + flow-summary.csv | ### Locked decisions - **SKU format**: `SQ8-{LINE}-{MIX}-{CONTAINER}-{SIZE}` — `SQ8-S-NE-CBD-2OZ` etc. No year in SKU (volatile data lives in price-history + cost-batch tables). - **Margin tiers**: 30% loss-leader (CBD-2OZ), 65% Xerces premium (XCC/XCV), 55% standard. - **Min-stock thresholds**: 5 lb for regular mixes, 2 lb for specialty (XCC, XCV, FL, NV, TX). - **Retail formula**: `retail = ceil_to_99((fixed_tier + container_$ + seed_$) ÷ (1 − target_margin))`. - **Free-shipping-embedded retail**: USPS cost lives in COGS, retail advertised "free shipping." - **Bag tier deferred to v3.12**: 64 SKUs at $19 placeholder until reformulated. --- ## 3 · The 30-day roadmap (in priority order) ### NOW · this week **N1. Operator's first Monday-morning use** - Visit `wildfag.com/purchasing.html`, confirm the top-of-list (NE) shows 0 buy-today (because 2× 5lb already on order) - Check that velocity numbers match operator gut-feel - If wrong → tell me which mix and what the right number is, I'll trace it back to the data - Pass: operator opens the page Monday and uses it as the source of truth for that week's POs **N2. Cost refinement for honest margin** (bag tier especially) - Current bag-tier COGS = `$7.50 fixed + seed_per_lb × oz / 16` — placeholder, probably understates real cost - Operator provides: actual pouch cost, label cost, USPS rate by weight, labor floor for bags - I update `cogs-matrix.csv` formula for bag tier - Margin numbers across all dashboards become trustworthy (right now they're labeled "estimated") **N3. Git repo + GitHub remote** - User runs `git init` at `scripts 2026/` root (commands ready, sandbox can't do it) - Push to `github.com//code-compendium-26` (private repo) - Daily commits going forward; `git ftp push` for one-command deploys (later) ### SOON · next 2–4 weeks **S1. Master Batch tracker (Notion integration)** - Operator creates Notion DB "Master Batches" with: batch ID, mixes×shakers, target qty, packed qty, status, owner, ship date - Share DB with Cowork Notion MCP integration - I build a Notion → batch-board renderer (scaffold exists in `05-03 SFIB Ops Dashboard/PLAN.md`) - New page: `wildfag.com/batches.html` — open batches, sub-batches cut, assembler leaderboard **S2. Real SLA tracking** - Fetch `sfinbloom.art/shipping-return-policy` → parse handling time + carrier promise - Compute per-order: days_open vs SLA threshold - Late-orders queue surfaces in Monday Morning + a dedicated `late-orders.html` - Signal: which orders need customer-service follow-up *today* **S3. Push the SQ8- pricing to the live store** - Single-row dry-run already passed (BAY-CBD-2OZ) - Full 144-row push: `wildfag-local/`-style audit page reviewed, then SQ admin → Import → upload - After: re-export, drop into `sq-import-audit.html` to confirm zero diff vs intended state - Decision pending: do we push at the operator's chosen time (current live prices are below formula on 4 of 4 mapped shakers, so this raises prices) **S4. Per-mix cover-weeks override** - Currently `purchasing.html` uses one cover-weeks setting for all mixes - Add per-mix override: e.g. NE/MW/SE = 6 weeks (top sellers, more buffer), specialty = 3 weeks - Stored in `inputs/policy/cover-weeks.json` ### LATER · next 1–3 months **L1. Tracking number → live ETA** - Operator pastes tracking number into incoming-shipments.csv - Daily script polls carrier API (USPS / UPS / FedEx) for updated ETA - Falls back to the manual `expected_arrival` if no tracking known - Dashboards update automatically without operator touch **L2. Forecasting layer** - Ingest 4+ weeks of order data (we currently have 2 months of sample) - Compute: lb/wk velocity, weeks-of-cover, projected stockout date - Seasonal lift modeling (May–Aug peak) - Reorder trigger: `weeks_cover < lead_time_days/7 + safety_buffer` → flag amber even if currently above min **L3. Throughput model from Clockify + ShipStation** - Days-to-pack = orders_per_day ÷ avg_pack_time (Clockify) - Days-to-ship = packs_per_day ÷ shipstation_throughput - Live "pulse" — combined with reorder ETA → customer-service can give real ETAs - Customer-facing impact: order page can quote "ships in 2 days" instead of "1–3 weeks" **L4. Bag tier (v3.12) launch** - Reformulate bag-tier COGS with real costs (see N2) - Lock retail prices - Generate 64 new SQ8-B-* SKUs in cogs-matrix - Update `recipe-matrix.html`, `pricing-table.html` to surface bag prices - Push to SQ as part of a coordinated launch announcement **L5. Cloudflare Access (AI agent + team)** - Replaces basic auth with team-managed access - Service tokens for AI / scripts to fetch live pages - Free tier covers up to 50 users - Lets me actually see how the live site renders (currently I work from operator screenshots) ### PARKED / discoverable - `seed-distributor.html` — old idea, mostly subsumed by `purchasing.html` (same math, different UI) - `pricing-lock-v2.html` live-prices toggle — `pricing-table.html` does this better - ShipStation API integration (currently CSV-drop) - SquareSpace API integration (currently CSV-drop) - Markdown wrapper for `*.md` files served from wildfag.com - Multi-channel SKU bookkeeping (only matters if SFIB ever sells same SKU on Amazon/wholesale at different prices) --- ## 4 · Known gaps in current state | Gap | Impact | Workaround | |---|---|---| | Bag-tier COGS is rough | Margin estimates on 4oz/8oz/1lb bags are unreliable | Numbers labeled "est. gross" in dashboards; operator instinct overrides | | SLA threshold hardcoded to 3 business days | Late-order count may be off vs. real shipping promise | Operator can mentally adjust until policy is fetched | | Order CSV is from one historical export | Velocity numbers reflect Mar 7 – May 3 only | Re-export weekly, drop into `reference/`, re-run `orders-to-flow.py` | | 12 unmapped order rows in flow-anomalies.txt | Tiny revenue blind spot | Add to variant_key.csv as legacy SKUs surface | | Margin uses formula COGS, not historical cost-at-time-of-sale | Historical orders shipped under different supplier prices than current | Use as estimate; refine when L1 (real cost-batch table) lands | | `index.html` still says "password-gated" / "wildfag-local" in tagline | Cosmetic — page is gated, but copy reads stale | Single-line edit when noticed | --- ## 5 · The systems map (where data flows) ``` SUPPLIERS SHIPSTATION SQUARESPACE │ │ │ ▼ ▼ ▼ pcs-invoices.csv orders.csv (export) products.csv (export) am-invoices.eml (parsed) ↓ ↓ incoming-shipments.csv orders-to-flow.py sq-import-audit.html │ │ │ └──────────┬───────────────────────┤ │ │ │ │ ▼ ▼ ▼ cogs-matrix.csv flow-data.json squarespace-import-shakers.csv (locked formula) (timeseries data) (144-row SQ8 import) │ │ │ └──┬────────────────────┴──────────┬───────────────────┘ │ │ ▼ ▼ recipe-matrix.html monday-morning.html pricing-table.html purchasing.html sfib-matrix.html inventory-flow.html pricing-lock-v2.html inventory-dashboard.html │ └────────► wildfag.com (deploy via Transmit) ``` --- ## 6 · Operator's daily ritual (what we're aiming for) ``` ☀ 8 am Monday 1. Open wildfag.com/monday-morning.html → see week's top sellers + critical alerts 2. Open wildfag.com/purchasing.html → see "buy today" total at top, hit suppliers for any non-zero amounts 3. Open Notion → Master Batches → mark this week's planned batches → those flow through to wildfag.com/batches.html for the team 4. Done before coffee finishes brewing ``` Currently we're at step 2 (purchasing live today). Step 3 needs the Notion DB. Step 1 needs SLA tuning. --- ## 7 · Hand-off checklist for any future helper If someone (AI or human) is picking this up cold: 1. Read `COMPENDIUM.md` for folder layout 2. Read THIS file for project state 3. Spot-check `cogs-matrix.csv` to see locked formula values 4. Spot-check `variant_key.csv` to see SKU mapping 5. Open one of the live dashboards to see the render 6. Run `scripts/orders-to-flow.py` to confirm pipeline still works against fresh CSVs 7. Don't fabricate product copy. Don't bulk-import without dry-runs. Don't put PII files in git. --- *This brief replaces the implicit chat-context spaghetti. Update it whenever a major decision lands; keep it short.*