Skip to main content
Glama
h9-tec

model-family-tree

by h9-tec
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault

No arguments

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server publishes no instructions, or was last inspected before Glama recorded them.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{
  "listChanged": false
}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}

Tools

Functions exposed to the LLM to take actions

NameDescription
lineageA

Ancestry of one Hugging Face model: the chain of declared base models up to the root, plus how many descendants the model itself has.

Walks `base_model` metadata live from the Hub (first parent at each step; merges list all
parents in `parents`). Each hop carries the relation (finetune, adapter, quantized, merge)
and the parent's descendant counts from the nightly build.

Example: lineage(id="unsloth/Qwen2.5-7B-Instruct")
childrenB

Direct descendants of a model, most downloaded first (or sort="likes").

`relation` filters to finetune, adapter, quantized or merge. Live from the Hub API;
falls back to the nightly top-25 list when offline. Example: children(id="Qwen/Qwen3-8B", relation="finetune")
leaderboardB

Most-descended base models on the Hub.

kind: all_time (unique descendants of every kind), 90d (new descendants in the last 90
days), finetune (finetune descendants only; filters out LoRA and quantization noise),
families (roots grouped by series across orgs), orgs (by organization owning the roots).
familyC

One model family (e.g. "Qwen2.5", "Llama 3.1", "FLUX.1"): totals, relation breakdown, roots, organizations and the top models in it by descendants.

searchA

Find models with descendants whose id contains the query (case-insensitive), most descended first. Covers the 88k models that have at least one declared child.

statsB

Headline numbers for the current build: models tracked, links, parents, roots, families, dates.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A3.5/5.0

Scored across 6 tools

Disambiguation4/5

The core directional tools are clearly distinct: lineage walks ancestry upward while children lists descendants downward, and search/stats are unambiguous. family vs leaderboard have some conceptual overlap (family gives a single family's totals/roots/orgs while leaderboard ranks families), but the descriptions differentiate them adequately.

Naming Consistency5/5

All six tools use a single, lowercase noun (family, lineage, children, leaderboard, search, stats), with no camelCase/snake_case mixing or verb-noun deviations. The convention is uniform and predictable across the whole set.

Tool Count5/5

Six tools is well within the sweet spot for this domain, and each covers a distinct query type (single-family, ancestry, descendants, rankings, search, aggregate stats). Nothing feels redundant or missing in terms of count.

Completeness4/5

The surface covers the full genealogy lifecycle: upward ancestry, downward descendants, per-family breakdowns, rankings, search, and build stats. Minor gaps exist (e.g. no direct relation-type aggregation or merge-graph view beyond lineage), but core workflows are well served.

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources