v0.5.1

Audit release.

• MCP: the mcp package is now a core dependency, so uvx gnomancer serve works out of the box; registry icon fixed.
• Report: smarter at-a-glance (carrier states and high-odds sites get their own lines; PharmCAT genes that could not be resolved are no longer reported as normal; HLA-B*57:01 surfaced), sticky contents with top links, deep links open collapsed sections, phone-friendly tables and chart, new methods and glossary sections, tidy drug-gene tables, no more "nan%".
• Pipeline: hemizygous X calls in males are no longer flagged as artifacts; protective sites are not counted as negatives; robust yhaplo install under uv-created venvs; --names validated in family mode.
• Installer: correct Linux Python version check, Java 17 check, broken-venv repair, PATH instructions.
• Docs: NOTICE.md credits, concrete health-wording rules, Java 17 requirement stated everywhere.

gnomancer v0.5.0

• Family mode: gnomancer family dad.zip mom.zip me.zip (and the MCP tool family). Shared DNA segments per pair via apriha/lineage with a relationship estimate, shared maternal/paternal lines, side-by-side panel genotypes, which parent each allele came from for parent-parent-child sets, and a Mendelian-consistency check. Synthetic family generator for testing.
• Haplogroup guide pages on gnomancer.com, linked from the report's lineage cards
• Installer skips Pythons whose XML module is broken (Homebrew bottles on macOS 26.2) and can fall back to a uv-managed Python

gnomancer v0.4.2

• First run downloads the open-source tools and references automatically (CLI and the MCP analyze tool), so uvx gnomancer serve is a zero-step install for assistant users. --no-setup opts out.
• SHA-256 pinning for downloaded tools; ClinVar verified against NCBI's MD5
• Homebrew tap: brew install codebase/tap/gnomancer

gnomancer v0.4.1

• One-page pharmacogenomics card (pgx-card.pdf / .html) for a pharmacist or prescriber: PharmCAT diplotypes and phenotypes, the drugs each gene matters for, clotting and abacavir single-site results, and a clear 'array-based, provisional' footer. Linked from the report.
• Announcement drafts in docs/announcements.md

gnomancer v0.4.0

• Population frequencies on every panel row: how common your genotype is in each 1000 Genomes super-population (Hardy-Weinberg from Ensembl GRCh37 allele frequencies), shown for your ancestry group with all five on hover
• gnomancer watch REPORT_DIR: ClinVar reclassification check. Keeps a snapshot of every ClinVar-classified allele you carry, downloads new ClinVar releases, rescans, and reports only what changed. Also an MCP tool (clinvar_watch).
• Protective homozygotes now listed under Reassuring; realistic synthetic demo profile
• Discord community: https://discord.gg/aGjqzqErQH

gnomancer v0.3.2

Registry metadata: MCP server name com.gnomancer/gnomancer. Installer now uses the PyPI package.

gnomancer v0.3.1

First release on PyPI: pipx install gnomancer && gnomancer setup.

• Package data ships inside the wheel; downloaded tools live in ~/.gnomancer via gnomancer setup
• MCP server (gnomancer serve), Claude skill, registry manifest
• Trusted publishing from GitHub Actions

gnomancer v0.3.0

• MCP server: gnomancer serve (stdio). Tools: analyze, summary, section, lookup, narrative_prompt, set_narrative, clinvar_findings, pharmacogenomics; resources for the get-your-data guide and the current report. Works with Claude Desktop, Claude Code, Cursor and any MCP client, entirely locally.
• Registry manifest (server.json) and publishing notes (docs/mcp-registry.md)
• Claude skill and plugin manifest
• gnomancer.com: get-your-data and install guides, one-line installer

gnomancer v0.2.0

• Plain-language 'at a glance' digest and a narrative-summary slot (--summary, summary-prompt.md, or --ai via Claude)
• ClinVar scan of every typed site with plausibility scoring
• PharmCAT star-allele pharmacogenomics with drug context and per-gene coverage
• Yleaf second-opinion Y haplogroup on current trees; ezancestry population estimate
• Editorial pass: readable condition and population names, no placeholder readings

gnomancer v0.1.0

First public release.

• Parses AncestryDNA, 23andMe, MyHeritage, FamilyTreeDNA, LivingDNA and VCF exports (via snps), remaps to GRCh37
• Quality metrics, independent sex inference, runs of homozygosity
• Y haplogroup via yhaplo, with a second opinion from Yleaf on current YFull / FTDNA / ISOGG trees
• mtDNA haplogroup via Haplogrep3 against rCRS
• Curated 111-site marker panel (ancestry, traits, health, carrier, pharmacogenomics) with strand checking
• ClinVar scan of every typed site, with plausibility scoring for array artifacts
• PharmCAT star-allele pharmacogenomics with per-gene coverage
• Optional ezancestry population classification
• Self-contained HTML report (light/dark, print, search, health sections collapsed by default) plus Markdown, TSV and JSON

Everything runs locally; the only network traffic is downloading public tools and references.
Not medical advice: confirm anything that matters with a clinical-grade test.