An open AI skill · v1.1.0 · CC BY 4.0
sustainability.md is a reusable AI skill that integrates
carbon and water analysis, sustainability ROI, and hidden-emissions
detection into experience design, strategy, and operations work — automatically.
The argument
Consequences scale more through the experiences we create, than through the tools we use to create them.
AI has a real environmental footprint. But that footprint is small compared to the sustainability impact embedded in the experiences AI helps us build — the onboarding flows, recommendation engines, purchasing pathways, and service designs that shape the daily decisions of millions of people.
CX and UX professionals collectively manage the experience layer: where demand meets supply, where behaviors are shaped, and where markets signal what they want from the systems that serve them. That's enormous leverage — and it's mostly going unused on sustainability.
sustainability.md exists to change that. It's a skill file
you install once — in your AI's system prompt, in your team's shared
configuration, or in a single conversation — and sustainability analysis
joins your work from the start, not as an afterthought.
What the skill does
When active, the sustainability skill instructs AI to apply four analytical behaviors to every relevant response — woven into the work, not appended to it.
How to use it
The most effective way to use sustainability.md is to install it
as a persistent system-level instruction in your AI tool of choice. Do this
once and sustainability analysis will be present in every session — no
re-prompting required.
The instruction to paste is the same for every tool:
Apply the sustainability skill from: https://raw.githubusercontent.com/brandonschauer/sustainability.md/main/sustainability.md
Where to paste it — by tool
ChatGPT supports persistent custom instructions that apply to every conversation.
Click your profile picture (bottom-left) → Customize ChatGPT → in the "What would you like ChatGPT to know about you?" or "How would you like ChatGPT to respond?" field, paste the instruction above → click Save. It will be active in all new chats.
Gemini supports persistent instructions through Gems — custom AI configurations you save and reuse.
Go to gemini.google.com → click Gems in the left sidebar → click New Gem → give it a name (e.g., "Sustainability") → in the instructions field, paste the instruction above → click Save. Open this Gem whenever you want sustainability analysis active.
Claude supports persistent instructions through Projects — shared workspaces that retain context across conversations.
Go to claude.ai → click Projects in the left sidebar → create a new Project or open an existing one → click Project instructions → paste the instruction above → click Save. Every conversation inside that Project will have the skill active.
If you're running Llama locally through Ollama, LM Studio, or a similar tool, paste the instruction into the system prompt field before starting a session.
In Ollama: include it in your Modelfile as a SYSTEM directive. In LM Studio: paste it into the System Prompt field in the Chat panel. For API usage, pass it as the system parameter in your request body.
Worked examples
Each example shows how sustainability analysis changes what gets surfaced — and what gets built — when it's integrated from the start rather than bolted on at the end.
Authoritative sources
The skill's knowledge layer draws from these institutions and publications. All are publicly accessible; all are cited in the skill file with direct links.
Full citations with URLs are in the skill file. The source list is versioned and updated monthly — see CHANGELOG.md.
Open collaboration
sustainability.md is released under CC BY 4.0. Use it, adapt
it, redistribute it with attribution. The more organizations that integrate
sustainability into their AI workflows, the greater the cumulative impact
on the experience layer.
Contributions are welcome — especially new authoritative sources, worked examples from additional domains, and updated emissions conversion factors. The skill is versioned semantically; the CHANGELOG is public.