Dog Geroscience

dog-geroscience-mcp

mcp-name: io.github.w0lph/dog-geroscience-mcp

An MCP server that gives any agent (Claude, Cursor, other MCP clients) dog-specific aging tools that do not exist anywhere else:

Tool	Source	What it answers
anage_species	HAGR AnAge	Longevity record for the dog (or any species for comparison): max longevity, IMR, MRDT, weight, maturity.
drugage_search, drugage_species_summary	HAGR DrugAge	Lifespan-extension experiments by compound/species, with the NIA ITP flag; shows how thin the dog evidence is.
genage_search	HAGR GenAge	Human aging genes and model-organism longevity genes by symbol/name.
dog_ortholog	Ensembl Compara (live, cached)	Dog ortholog(s) of a human/mouse gene: id, symbol, % identity, orthology type, location.
dose_translate	FDA 2005 Km table	mg/kg conversion between species (mouse → dog = ×3/20).
dap_releases, dap_codebook_search, dap_variable	Dog Aging Project public codebooks (GitHub)	Which survey variables capture a concept; full entry for a variable; which releases contain it.
nih_reporter_search	NIH RePORTER v2 (live)	Funded projects matching a phrase, by fiscal year.
corpus_search, corpus_record, corpus_info	canine-aging-corpus (Europe PMC)	BM25 search over 3,500+ canine aging papers; abstracts for all, Markdown full text for the open-access subset.
foi_summary_search, foi_summary_get	foi-summaries (FDA CVM, public domain)	FOI summaries for 497 dog products: recommended dosage, indications, route, PK and safety candidate sentences, PDF link; foi_summary_get returns the parsed section text (e.g. the dose-multiple target-animal-safety study) by foi_id.
foi_structured_search	foi-summaries structured layer	Typed, quote-grounded records for all 497 summaries (1,026 PK values, 193 safety studies, 918 adverse-reaction rows): dose regimen, PK values, target-animal-safety design/findings, effectiveness study, adverse reactions with treated/control rates.
canine_trial_search	canine-trials registry	Interventional studies and cohorts on aging in dogs (lifespan, healthspan, cognition, mobility, immunosenescence, cancer prevention), including company programs (STAY/LOY-002, LOY-001, LOY-003) and TRIAD: design, status, intervention, dose, duration, population, primary outcome, result as reported, sources with verbatim quotes. Match on name, acronym, intervention, tags or PMID; trial_id for one record; status filter.
intervention_dossier	composition of the above + openFDA	One evidence package per compound: DrugAge across species with ITP and dog rows, dog-equivalent doses, corpus hits and dog trials, target ortholog, openFDA dog adverse events, FOI summaries, explicit gaps. Synonym-aware (selegiline ↔ L-deprenyl, rapamycin ↔ sirolimus).

The server also exposes one MCP prompt, dossier_briefing(compound, target_gene?), which instructs the client model to call intervention_dossier and write a six-section evidence briefing without adding anything the tools did not return. examples/rapamycin_briefing.md is a worked example written from examples/rapamycin_dossier.json; examples/selegiline_dossier.json shows the FOI and structured blocks populated for a marketed veterinary drug.

scripts/dossier_eval.py runs the dossier over 14 ITP-tested compounds plus two veterinary comparators and prints a coverage table (data/dossier_eval.json). It measures what the sources contain, not biological truth: for example rapamycin resolves 37 DrugAge experiments, 16 ITP rows, 152 corpus mentions and 5 candidate dog trials but no FOI summary (not a veterinary product), while selegiline resolves 1 dog lifespan experiment, 4,117 openFDA dog reports and 2 FOI summaries with 2 structured records, and carprofen 46,859 openFDA dog reports plus FOI summaries and structured records up to each search's result cap (11 and 19).

Adjacent servers this one deliberately does not duplicate: sniff-mcp (canine genomics, OMIA, breed allele frequencies), mcp-veterinary-fda (openFDA animal adverse events, Green Book), and the longevity-genie servers (Open Genes, SynergyAge, gget).

Install

No build step: the first run downloads the prebuilt database (~90 MB) from the Hugging Face Hub (w0lph/dog-geroscience-mcp-data) into a per-user cache directory, and every offline tool works from that single file.

uvx dog-geroscience-mcp                 # stdio server; downloads the database on first start
uvx dog-geroscience-mcp fetch-data      # or download it explicitly (prints the path)

Claude Desktop / Claude Code / Cursor config:

{
  "mcpServers": {
    "dog-geroscience": {
      "command": "uvx",
      "args": ["dog-geroscience-mcp"]
    }
  }
}

Claude Desktop, one click: download dog-geroscience-mcp-<version>.mcpb from the releases page and open it (Settings → Extensions → Install from file). The bundle declares the PyPI package as a uv-type extension, so the desktop app installs it with its own uv; nothing else to set up. Source of the bundle: mcpb/ (npx @anthropic-ai/mcpb pack mcpb dist/dog-geroscience-mcp-0.1.1.mcpb).

Claude Code, as a plugin with a routing skill (needs uv on the PATH):

/plugin marketplace add w0lph/k9
/plugin install dog-geroscience@k9

Codex CLI:

codex mcp add dog-geroscience -- uvx dog-geroscience-mcp

The skill alone, for any agent that reads SKILL.md files: copy plugins/dog-geroscience/skills/dog-geroscience/ into your skills directory (for Claude Code, ~/.claude/skills/). It routes dog-aging questions to the right tools and states how to read their output (species labels on FOI summaries, allometric doses, verbatim quotes).

Docker (the root Dockerfile of the repository bakes the database into the image):

docker build -t dog-geroscience-mcp . && docker run -i --rm dog-geroscience-mcp

Environment: DOG_GERO_DATA (where the database lives; default %LOCALAPPDATA%\dog-geroscience-mcp on Windows, ~/.cache/dog-geroscience-mcp elsewhere, <checkout>/data in a source checkout), DOG_GERO_DB_URL (alternative download URL), DOG_GERO_AUTO_FETCH=0 (fail instead of downloading when the database is missing).

Build from sources

uv sync
uv run dog-geroscience-mcp build            # downloads HAGR + DAP codebooks, indexes ../corpus/data and ../foi/data
uv run dog-geroscience-mcp build --skip-download
uv run dog-geroscience-mcp build --fulltext-licences "cc by,cc0" --out dog_geroscience.sqlite   # the redistributable build
uv run dog-geroscience-mcp                  # serve from data/dog_geroscience.sqlite
uv run mcp dev src/dog_geroscience_mcp/server.py   # inspector (optional)

The build writes one SQLite file: plain tables + FTS5 indexes, the corpus's Markdown full text (all of it locally; only CC BY / CC0 articles in the published file), the full FOI records, and a meta table with every source URL and fetch time. DOG_GERO_CORPUS, DOG_GERO_FOI and DOG_GERO_FOI_STRUCTURED point the build at the inputs.

In a source checkout, Claude Desktop can run it without PyPI:

{
  "mcpServers": {
    "dog-geroscience": {
      "command": "uv",
      "args": ["--directory", "D:/OneDrive/k9/mcp", "run", "dog-geroscience-mcp"]
    }
  }
}

Design

• Ground-truth tables (AnAge, DrugAge, GenAge, DAP codebooks) are served verbatim with their source; nothing is merged or inferred.
• Network at request time is limited to the live tools (Ensembl, RePORTER, and openFDA inside the dossier); Ensembl results are cached in the database. The database itself is fetched once, atomically, and checked for the SQLite header before use.
• Query functions in queries.py are pure and tested offline; server.py is a thin wrapper.
• Licences: code MIT; HAGR data CC BY 3.0 (commercial use permitted with attribution); DAP codebooks are public GitHub files; corpus content keeps each article's own licence.

Test

uv run pytest -q
uv run ruff check src tests

Tests build a miniature database from fixtures and drive the server through the SDK's in-process client; nothing touches the network.