ai-impact-ecologits

Estimates the environmental impact of AI/LLM sessions using the EcoLogits API.
Auto-detects GitHub Copilot CLI session tokens, or accepts manual token counts for any LLM tool.

---

What it does

Ask the agent about the environmental cost of your current session (or any time window) and it will report:

• ⚡ Energy use (kWh range)
• 🌱 CO₂ equivalent (GWP range)
• 🚗 Car distance equivalent (configurable for your vehicle and region)
• 💻 Workstation energy (based on your machine's wattage)
• 📊 Per-model breakdown when multiple models were used

Works with Claude, GPT, and any model in the map — across single sessions or aggregated over days/date ranges.

---

Installation

This is a skill for GitHub Copilot CLI. Install it with:

gh copilot skill install https://github.com/larry-oates/ai-impact-ecologits

Or using npx skills:

npx skills add larry-oates/ai-impact-ecologits

No other dependencies required — Node.js is bundled with Copilot CLI.

---

Usage

Once installed, just ask Copilot CLI naturally:

"What's the environmental impact of this session?"
"How much CO₂ has my AI usage generated this week?"
"Estimate the carbon footprint of this session for a user in Germany."

The skill triggers automatically on questions about environmental impact, carbon footprint, energy usage, CO₂ emissions, sustainability, or ecological cost.

---

Manual use

You can also run the script directly:

# Current Copilot CLI session (auto-detected)
node "./run-impact.js"

# Any LLM tool — pass your output token count manually
node "./run-impact.js" --Model gpt-5.4 --OutputTokens 1500
node "./run-impact.js" --Model claude-sonnet-4.6 --OutputTokens 800

---

Options

Time period

Parameter	Example	Description
--Days	--Days 7	Last N days (rolling window)
--Since	--Since 2025-04-01	From this date to now
--Until	--Until 2025-04-30	Up to this date

Combine --Since and --Until for a precise date range:

node "./run-impact.js" --Since 2025-04-01 --Until 2025-04-30

Regional & vehicle settings

Parameter	Default	Description
--Zone	WOR	Electricity mix — ISO 3166-1 alpha-3 country code or WOR (world average). E.g. USA, DEU, FRA, GBR, AUS, JPN, IND, BRA. Full list
--CarEfficiency	40	Your car's fuel efficiency (unit set by --CarUnit)
--CarUnit	mpg-uk	mpg-uk · mpg-us · kpl · l100km
--FuelType	Diesel	Diesel, Petrol, or Gasoline
--WorkstationWatts	150	Workstation power draw in watts

Examples by region

# USA
node "./run-impact.js" --Zone USA --CarEfficiency 30 --CarUnit mpg-us --FuelType Gasoline

# Europe (Germany)
node "./run-impact.js" --Zone DEU --CarEfficiency 6.5 --CarUnit l100km

# Australia
node "./run-impact.js" --Zone AUS --CarEfficiency 12 --CarUnit kpl

# UK
node "./run-impact.js" --Zone GBR --CarEfficiency 40 --CarUnit mpg-uk

Other options

Parameter	Default	Description
--Model	auto	Override the model name used for impact calculation
--OutputTokens	—	Skip session auto-detection; use this token count directly
--SessionId	auto	Override the auto-detected session UUID

---

Supported models

model-map.json maps model names to EcoLogits identifiers. 181 models are mapped across four providers:

Provider	Examples
Anthropic	claude-sonnet-4.6, claude-opus-4.6, claude-haiku-4.5, claude-3-haiku-20240307, dated variants
OpenAI	gpt-5.4, gpt-5.2, gpt-4.1, gpt-4o, gpt-4-turbo, gpt-3.5-turbo, o1, o3-mini, o4-mini, dated variants
Mistral AI	mistral-large-latest, mistral-medium-latest, codestral-latest, devstral-latest, magistral-small-latest, pixtral-large-latest, voxtral-mini-latest, and more
Cohere	command-r, command-a-03-2025, c4ai-aya-expanse-32b, and more

For Copilot CLI models the map translates dot-notation names (e.g. claude-sonnet-4.6) to the EcoLogits dash-notation equivalent. For all other tools, pass the model name exactly as it appears in the EcoLogits API.

Adding a model

If the script warns that a model isn't found, add it to model-map.json:

{ "copilot_model": "your-model-name", "provider": "anthropic", "ecologits_model": "ecologits-name" }

Check available EcoLogits model names at:
https://api.ecologits.ai/v1beta/models/{provider}

---

Powered by

• EcoLogits — environmental impact estimates for LLM inference
• GitHub Copilot CLI

---

License

MIT