Readmygenome

readmygenome

*A Claude skill that lets AI read `.genome` bundles correctly.*

Drop `readmygenome.md` into Claude's skills folder. Upload a `.genome/1.0` bundle. Ask anything.

you: What actionable findings do I have?
claude: [validates the bundle, runs the canonical query, returns findings
         with gene, HGVSp, ClinVar significance, and review stars — citing
         the exact is_actionable rule from the bundle's manifest]

What this is

`.genome` is an open specification for a consumer genome file designed to be read by an AI ([github.com/genome-computer/genome-spec](https://github.com/genome-computer/genome-spec)). This skill is the operational integration that lets Claude reason over a `.genome` bundle without re-introducing the failure modes the bundle was designed to prevent.

The skill is deliberately narrow. It answers factual questions about a bundle's contents, with provenance, against a versioned schema. It does not provide medical advice, drug recommendations, or diagnostic conclusions.

Install

**One-line install** (from the repository root):

cp readmygenome.md ~/.claude/skills/readmygenome/SKILL.md
cp -r scripts reference requirements.txt ~/.claude/skills/readmygenome/

Claude's skill loader looks for a file named `SKILL.md` inside each skill directory — the file in this repo is named `readmygenome.md` because that is the product name you'll see everywhere else. The rename is a one-line copy.

Python dependencies for the helper scripts:

pip install -r requirements.txt

That installs `duckdb`, `pyarrow`, and `jsonschema`.

Verify the install

Open a Claude conversation and type:

*What skills do you have loaded?*

You should see `readmygenome` in the list. If not, check that `~/.claude/skills/readmygenome/SKILL.md` exists and that its YAML frontmatter contains `name: readmygenome`.

Using with other agent systems

The heart of the skill is `readmygenome.md` — a single markdown file with YAML fron