📚Public vCon Datasets

Public vCon datasets on GitHub, with counts, and how to load one into a vcon-mcp environment.

vcon-dev publishes several public datasets as GitHub repos, each vCons in IETF vCon syntax 0.4.0. Some also run as a hosted, read-only MCP server for querying the data directly.

Datasets

Name	What it is	Count	Spec	Repo	Hosted endpoint
vCon Supreme Court Arguments	US Supreme Court oral arguments, terms 1955-2025	8,503	0.4.0	vcon-dev/vcon-supreme-court-arguments	mcp-scotus-open.vconic.com
IETF Meeting vCons	IETF working-group sessions, meetings 66-126 plus interims	8,181	0.4.0	vcon-dev/ietf-meeting-vcons	-
vCon Dataset: City of Newport, RI	City of Newport, RI public meetings	115	0.4.0	vcon-dev/vcon-dataset-city-of-newport-ri	-
Fake vCons	Synthetic customer-service vCons from vcon_faker	601	0.4.0	vcon-dev/fake-vcons	-
TADHack 2025	Synthetic TADHack 2025 demo calls with audio	43	0.4.0	vcon-dev/tadhack-2025	-

The SCOTUS and IETF datasets link to external audio; those external links do not carry a content_hash yet (work in progress). The Newport dataset's external audio links don't either, since the host blocks automated fetch.

Load one

vcon-data (package @vconic/vcon-data 0.1.0, repo VCONIC/vconic-datasets) is the CLI for deploying a dataset into a vcon-mcp Supabase environment. Install it:

npm install -g github:VCONIC/vconic-datasets

Set up an environment in ~/.config/vcon-data/envs.json (see the vcon-data README), then deploy a dataset straight from its GitHub repo:

vcon-data deploy github.com/vcon-dev/fake-vcons --to sandbox

deploy lints the dataset first and fails on any lint error. To check a local checkout on its own:

vcon-data lint ./my-dataset --json