Skip to content

charlieviettq/awesome-agent-skill / .claude/skills/biz-supply-chain

"biz-supply-chain"

Verified

"Analyze supply chain operations using the SCOR model across Plan, Source, Make, Deliver, and Return processes. Use this skill when the user needs to optimize supply chain efficiency, evaluate supplier performance, improve logistics, or design an end-to-end supply chain strategy — even if they say 'our deliveries are slow', 'supply chain costs are too high', or 'we keep running out of stock'.".

FileKindSize
SKILL.mdskill md5.8 KB
examples/sample_scenario.mdmarkdown5.4 KB
references/inventory-models.mdmarkdown9.7 KB
references/scor-metrics.mdmarkdown10.7 KB

references/inventory-models.md

View on GitHub
# Inventory Models Reference

Inventory sits at the intersection of **Plan** and **Source** in the SCOR model. Too much = cash tied up, obsolescence risk, high holding cost. Too little = stockouts, backorders, lost sales, and the downstream delivery failures that surface as customer complaints. These models give you the math to find the right amount.

---

## The Four Questions Every Model Answers

| Question | Model / Method |
|---|---|
| How much to order at once? | EOQ (Economic Order Quantity) |
| When to trigger a reorder? | ROP (Reorder Point) + Safety Stock |
| How much buffer to hold against uncertainty? | Safety Stock calculation |
| Which items deserve the most attention? | ABC / ABC-XYZ classification |

---

## 1. EOQ — Economic Order Quantity

EOQ minimizes total inventory cost by balancing ordering cost against holding cost.

### Formula

```
EOQ = √(2DS / H)

D = Annual demand (units/year)
S = Cost per order (setup/ordering cost, in $)
H = Annual holding cost per unit ($ per unit per year)
    H is often expressed as (unit cost × holding rate), e.g., $50 × 25% = $12.50/yr
```

### Worked Example

A Taiwanese electronics brand sells 12,000 units/year of a main PCB component.

- `D = 12,000` units/year
- `S = NT$800` per purchase order (admin, inspection, receiving labor)
- Unit cost = NT$500; holding rate = 25%/year
- `H = 500 × 0.25 = NT$125` per unit per year

```
EOQ = √(2 × 12,000 × 800 / 125)
    = √(19,200,000 / 125)
    = √153,600
    ≈ 392 units per order
```

**Number of orders per year** = 12,000 / 392 ≈ 31 orders
**Average cycle inventory** = 392 / 2 = 196 units

### Total Cost Check

```
Total cost = (D/Q) × S + (Q/2) × H
           = (12,000/392) × 800 + (392/2) × 125
           = 30.6 × 800  +  196 × 125
           = 24,490  +  24,500
           ≈ NT$48,990 / year
```

The symmetry (ordering cost ≈ holding cost) is a sanity check: at EOQ, these two costs are always equal.

### EOQ Assumptions and Where They Break

| Assumption | Reality Check |
|---|---|
| Demand is constant and known | Promotions, seasonality break this — see ABC-XYZ |
| Lead time is fixed and known | Supplier variability breaks this — add safety stock |
| No quantity discounts | If supplier offers tiered pricing, run EOQ at each tier and compare total cost |
| Infinite shelf life | Perishables require time-sensitive variants (e.g., newsvendor model) |
| Single item, single location | Multi-echelon inventory requires more complex models |

**Practical rule**: EOQ gives you the right *order of magnitude*. Don't treat the output as a precise target — round to the nearest pallet or MOQ, then recalculate total cost.

---

## 2. Reorder Point (ROP)

EOQ tells you *how much* to order. ROP tells you *when* to order.

### Formula

```
ROP = d̄ × L + SS

d̄ = Average daily demand (units/day)
L  = Lead time (days)
SS = Safety stock (units) — see Section 3
```

### Worked Example (continuing from above)

- `D = 12,000` units/year → `d̄ = 12,000 / 250 = 48` units/day (250 working days)
- Supplier lead time from Shenzhen: `L = 14` days
- Safety stock (calculated below): `SS = 180` units

```
ROP = 48 × 14 + 180
    = 672 + 180
    = 852 units
```

**Interpretation**: When on-hand inventory drops to 852 units, place the next order. It will arrive in ~14 days, just as stock approaches safety stock level.

---

## 3. Safety Stock

Safety stock is the buffer against two sources of variability: **demand uncertainty** and **lead time uncertainty**.

### Method A: Service Level Approach (Most Common)

```
SS = z × σ_demand × √L

z           = z-score for target service level
σ_demand    = Standard deviation of daily demand (units/day)
L           = Lead time (days)
```

**Service Level → z-score table:**

| Target Service Level | z-score |
|---|---|
| 90% | 1.28 |
| 95% | 1.65 |
| 97.5% | 1.96 |
| 99% | 2.33 |

Service level here = probability of *not* stocking out during a replenishment cycle.

### Method A Worked Example

Same PCB component:
- Target service level: 95% → `z = 1.65`
- Historical daily demand std dev: `σ_demand = 12` units/day
- Lead time: `L = 14` days

```
SS = 1.65 × 12 × √14
   = 1.65 × 12 × 3.742
   = 74.1 ≈ 75 units
```

This is safety stock against **demand variability only** (assumes lead time is constant).

### Method B: Combined Demand + Lead Time Variability

When the supplier's lead time is also variable (the Shenzhen supplier in the SKILL.md example):

```
SS = z × √(L × σ_d² + d̄² × σ_L²)

σ_d = Std dev of daily demand
σ_L = Std dev of lead time (days)
d̄   = Average daily demand
L   = Average lead time
```

### Method B Worked Example

- `z = 1.65` (95% service level)
- `d̄ = 48` units/day, `σ_d = 12` units/day
- `L = 14` days, `σ_L = 3` days (supplier on-time rate is inconsistent)

```
SS = 1.65 × √(14 × 12² + 48² × 3²)
   = 1.65 × √(14 × 144 + 2304 × 9)
   = 1.65 × √(2,016 + 20,736)
   = 1.65 × √22,752
   = 1.65 × 150.8
   = 248.8 ≈ 249 units
```

Adding lead time variability more than **tripled** the required safety stock (75 → 249 units). This is why fixing the Shenzhen supplier's reliability (SCOR Source metric: Supplier On-Time Rate) directly reduces inventory cost — it's not just about delivery predictability, it cuts the safety stock requirement.

**Revised Inventory Days** with Method B:
- Cycle stock: 392 / 2 = 196 units
- Safety stock: 249 units
- Average total inventory: 196 + 249 = 445 units
- At 48 units/day → **9.3 days of inventory**

---

## 4. ABC Classification

Not all SKUs deserve the same inventory model and attention. ABC classifies by revenue or cost contribution.

### Classification Rules

| Class | % of SKUs | % of Revenue/Value | Treatment |
|---|---|---|---|
| A | ~20% | ~80% | Tight control, frequent review, sophisticated models |
| B | ~30% | ~15% | Moderate control, periodic review |
| C | ~50% | ~5% | Loose control, simplify, min-max policies |

### How to Run the Classification

```
1. List all SKUs with annual usage value = (annual units sold × unit cost)
2. Sort descending by annual usage value
3. Calculate cumulative % of total value
4. Assign A / B / C cutoffs (80% / 95% / 100% cumulative value)
```

### ABC-XYZ Extension

Add a second dimension for **demand variability** (XYZ):

| Class | Coefficient of Variation (CV = σ/μ) | Meaning |
|---|---|---|
| X | CV < 0.5 | Stable, predictable demand |
| Y | 0.5 ≤ CV < 1.0 | Seasonal or moderate variability |
| Z | CV ≥ 1.0 | Highly irregular, lumpy demand |

**CV formula:**
```
CV = (Standard Deviation of demand) / (Mean demand)
     measured over rolling 12 months, same time bucket (weekly or monthly)
```

### ABC-XYZ Decision Matrix

| Class | Inventory Approach |
|---|---|
| AX | EOQ + ROP with tight safety stock; review weekly |
| AY | Safety stock + S&OP integration; review weekly |
| AZ | Make-to-order if possible; no bulk inventory; escalate to Plan team |
| BX | Standard ROP; review monthly |
| BY | Min-max policy; review monthly |
| BZ | Min-max with buffer; consider consolidating with suppliers |
| CX | Min-max with large lot (reduce ordering cost); review quarterly |
| CY | Periodic review; consider consignment from supplier |
| CZ | Eliminate SKU or special-order only; high risk of obsolescence |

**AZ items are the supply chain manager's biggest headache**: high value AND unpredictable. Do not apply EOQ here — it assumes stable demand and will systematically under- or over-stock.

---

## 5. Inventory Days (DSI) and Target-Setting

The SKILL.md SCOR dashboard uses **Inventory Days** as the Plan process metric.

```
Inventory Days (DSI) = Inventory / (COGS / 365)

or equivalently:

DSI = Average Inventory (units) / Average Daily Demand (units/day)
```

### Industry Benchmarks

| Industry | Typical DSI Target |
|---|---|
| FMCG / Grocery | 15–30 days |
| Consumer Electronics | 30–60 days |
| Automotive Parts | 20–45 days |
| Apparel (fast fashion) | 30–60 days |
| Industrial B2B | 45–90 days |
| Pharmaceuticals | 60–120 days |

**Warning**: Low DSI is not always good. If DSI drops below lead time + safety stock coverage, stockouts are guaranteed. Calculate the floor:

```
Minimum viable DSI = Lead Time (days) + Safety Stock Days
                   = L + (SS / d̄)
```

Using the example above: `14 + (249 / 48) = 14 + 5.2 ≈ 20 days minimum`.
Pushing DSI below 20 days for this item would cause stockouts even with 95% service level target.

---

## 6. Choosing the Right Model — Decision Tree

```
Is demand stable and continuous?
├── YES → Use EOQ + ROP
│         ├── Is lead time variable?  → Use Method B safety stock
│         └── Is lead time fixed?     → Use Method A safety stock
│
└── NO → Is it lumpy / project-based?
          ├── YES → Make-to-order; no standing inventory
          └── NO (seasonal) → Use time-phased MRP or newsvendor model
                              (outside scope of this reference)
```

**Additional forcing conditions:**
- Supplier requires MOQ > EOQ → Order at MOQ, recalculate cost
- Product has shelf life < lead time → Cannot pre-stock; redesign supply chain
- Demand CV > 1.0 (Z class) → Standard EOQ/ROP will fail; escalate

---

## 7. Interaction with SCOR Metrics

| Safety Stock Driver | SCOR Process | Metric to Improve |
|---|---|---|
| High `σ_demand` | Plan | Forecast Accuracy |
| High `σ_L` (supplier variability) | Source | Supplier On-Time Rate |
| Long `L` (average lead time) | Source | Supplier Lead Time |
| High `z` required (high service target) | Deliver | Perfect Order Rate target |
| High unit cost → high `H` | Source | Unit Cost / Make-vs-Buy |

The Iron Law applies directly here: **reducing safety stock requires fixing the upstream root cause** (forecast accuracy or supplier reliability), not just adjusting the inventory model parameters.
brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 323kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 198kUpdated Claude CodeCodexVerifiedInstall