More Than Human Skills — Development skill for Claude Code

More-Than-Human Skills

Agentic skills and commands for checking whether what we build serves all living things — not just humans.

**14 skills and 3 commands across 1 plugin.**

Why this exists

Design has a boundary baked into its vocabulary. A *user* is a human. A *stakeholder* is a human with budget. Everything else — the animals in the supply chain a recommender optimises, the watershed a datacentre draws from, the species whose habitat a mapping tool makes legible for clearing — is an externality by construction.

It isn't that teams decide against non-human life. It's that no step in the process asks. There is no slot in a design review, a PRD, or a launch checklist where a river or a barn owl could come up, so it doesn't.

This repo is an attempt at that slot.

It draws on work that already exists and is rarely applied in product teams: **more-than-human design** (Wakkary, Forlano), **animal-computer interaction** (Mancini), multispecies ethnography, and the environmental-computing literature on the energy, water, and material cost of machine learning.

What stops it being decoration

This genre collapses into virtue signalling in about a page. Three commitments hold it open:

**Name specific beings, not "nature."** "Migratory birds along the Atlantic flyway" is auditable. "The environment" isn't. Every stakeholder entry has to be specific enough that someone could go and check it.

**Quantify with honest proxies.** kWh, litres, hectares — stated as ranges, with the measurement boundary attached, and with uncertainty admitted rather than smoothed. Not a five-point vibes score.

**Force a decision.** Every finding ends in a change to the product, or it gets logged as accepted harm with a named individual attached. There is no third option and no composite score, because a score is how a decision gets made without anyone making it.

The four lenses

Lens	Question	Skills
Who is affected	Which living things does this to