run-lca

version	0.2
name	run-lca
author	Calvin Williamson (calvinw)
description	Use this skill whenever the user wants to perform a Life Cycle Assessment (LCA) using openLCA. Triggers include: "carbon footprint of", "compare environmental impact", "LCA for", "which material has lower impact", "run an LCA", "ecoinvent dataset for", "GWP of", "calculate impact", "environmental footprint", "lifecycle analysis", "hotspot analysis", "sensitivity analysis", "scenario analysis", "cradle to gate". Also triggers when the user asks to make a spreadsheet, slides, or report FROM an LCA result.

How this skill works

Call the run_lca MCP tool with the product graph YAML as a string. The tool returns LCI totals, LCIA impact scores, scaling vector, and two SVG supply chain diagrams — all in one call. No bash commands needed.

run_lca(product_graph="<yaml string>")

The tool returns:

{
  "name": "...",
  "method": "TRACI 2.2",
  "functional_unit": "1 shirt — One cotton t-shirt",
  "system_id": "...",
  "lci": {
    "Carbon dioxide": { "amount": 0.38, "unit": "kg", "type": "emission" },
    "Water":          { "amount": 2125, "unit": "L",  "type": "resource" }
  },
  "lcia": {
    "Global warming": { "score": 0.396, "unit": "kg CO2 eq" },
    "Acidification":  { "score": 0.0,   "unit": "kg SO2 eq" }
  },
  "scaling_vector": {
    "P1 — Cotton farming": 0.266,
    "P2 — Yarn spinning":  0.231
  },
  "svg_scaled":    "<svg>...</svg>",
  "svg_structure": "<svg>...</svg>"
}

Display both SVGs inline so the student can see the supply chain diagram. Explain the results in plain English — never show the raw JSON to the student.

---

Prerequisites

Check that the Life Cycle Assessment MCP server is running using the check_server tool before calling run_lca. If it returns {"running": false}, tell the student:

"The calculation engine isn't running right now. Start it by running cd mcp-lca && uv run python sse_server.py in the terminal, then try again."

---

product_graph.yaml format

The product graph is a YAML string (the frontmatter of a product_graph.yaml file). Pass it directly to run_lca — no file path needed.

name: <human name for this analysis>
goal: <one sentence describing the purpose>

functional_unit:
  description: <what is being analysed, in plain English>
  amount: 1.0
  unit: <unit symbol, must appear in 'units' below>

units:
  cup: Cup count
  L:   Volume
  MJ:  Energy
  kg:  Mass

products:
  - { name: <flow name>, unit: <symbol> }

elementary_flows:
  emissions:
    - { name: <flow name>, unit: <symbol> }
  resources:
    - { name: <flow name>, unit: <symbol> }

processes:
  - name: <process name>
    reference_output: { flow: <product name>, amount: <number> }
    inputs:
      - { flow: <product name>, amount: <number> }
    emissions:
      - { flow: <emission name>, amount: <number> }
    resources:
      - { flow: <resource name>, amount: <number> }

reference_process: <process name that delivers the functional unit>

lcia:
  method_name: "TRACI 2.2"

Minimal working example (coffee)

name: Coffee LCA — one cup
goal: Calculate CO₂ for one cup of coffee
functional_unit:
  description: One cup of coffee
  amount: 1.0
  unit: cup

units:
  cup: Cup count
  L:   Volume
  MJ:  Energy
  kg:  Mass

products:
  - { name: Coffee,       unit: cup }
  - { name: Boiled water, unit: L   }
  - { name: Electricity,  unit: MJ  }

elementary_flows:
  emissions:
    - { name: Carbon dioxide, unit: kg }

processes:
  - name: P1 — Make coffee
    reference_output: { flow: Coffee,       amount: 1.0 }
    inputs:
      - { flow: Boiled water, amount: 0.2 }

  - name: P2 — Boil water
    reference_output: { flow: Boiled water, amount: 1.0 }
    inputs:
      - { flow: Electricity,  amount: 0.5 }

  - name: P3 — Burn coal
    reference_output: { flow: Electricity,  amount: 1.0 }
    emissions:
      - { flow: Carbon dioxide, amount: 1.0 }

reference_process: P1 — Make coffee

lcia:
  method_name: "TRACI 2.2"

---

FEDEFL flow names

Always use exact EPA FEDEFL names for elementary flows:

Use this	Not this
Carbon dioxide	CO2, CO2 to air
Methane	CH4
Nitrous oxide	N2O
Nitrogen oxides	NOx
Sulfur dioxide	SO2
Water	H2O

---

What to ask the user when setting up a new analysis

1. Name and goal — what system, why are they studying it?
2. Functional unit — what, how much, which unit?
3. Units — list every unit symbol used.
4. Products — intermediate goods flowing between processes.
5. Emissions — CO₂, CH₄, NOₓ, freshwater, etc.
6. Processes — for each: what does it produce, consume, and emit?
7. Reference process — which process delivers the functional unit?

---

After getting results

Explain everything in plain English. Never show raw JSON. Always:

1. Show both SVG diagrams (scaled and structure)
2. Call out the global warming score first — it's the number students understand most easily
3. Explain the scaling vector — which processes ran how much?
4. Identify the hotspot — which process contributed most to GWP?
5. Connect the result to a real business decision

Example:

"To produce one cotton t-shirt, the cotton farm only needs to run at 27% capacity — but it still accounts for most of the water use (2,125 litres). The total climate impact is 0.40 kg of CO₂-equivalent, roughly the same as driving a car for 3 kilometres."