Track Hub issues with code and eval fixtures
Connect
SPIRITT
Hugging FaceHugging Face
Research into releases
SPIRITT works across Hugging Face models, datasets, and Spaces to move experiments into usable software. Curate evidence, maintain repositories, and operate the work around each release.
What you can do with Hugging Face.
Choose usable data
Inspect splits, rows, size, statistics, and dataset metadata before investing in a training or evaluation pipeline.
Maintain release context
Keep model files, repository discussions, version tags, and documentation aligned with the work your team ships.
Build beyond discovery
Turn chat completions and embeddings into practical applications, with Hub research and source revisions kept traceable.
Works with your other apps.
Get started in three steps.
01
Open a workspace
Create a SPIRITT workspace for the systems you want built and the recurring work you want owned.
02
Connect Hugging Face
Connect Hugging Face using account authorization or an API key with access to the repositories your mission needs.
03
Hand over a mission
Choose a dataset review, model release, or Space to build, and give SPIRITT the acceptance criteria and repository scope.
Hand over a mission.
Choose a mission. Make it yours.
Put Hugging Face to work.
Move your next model forward
Questions
Hugging Face, with SPIRITT.
01Which parts of Hugging Face can SPIRITT work with?+
It can inspect and maintain model, dataset, and Space repositories; work with branches, commits, and tags; manage discussions; and curate collections. Chat completions and text embeddings support applications beyond repository administration.
02How are private or gated repositories handled?+
The connected account and its authorization determine access. SPIRITT can request repository access and, for datasets you administer, list and handle user access requests or update gating settings according to your policy.
03Can it inspect a dataset without loading everything?+
Yes. It can list configurations and splits, retrieve first rows or selected rows, filter rows, and inspect sizes and statistics. Croissant metadata adds a structured description for checking whether a dataset suits the task.
04Can it publish and maintain a Space?+
SPIRITT can create a repository, commit Space files, manage variables and secrets, and create revision tags. Space information, events, metrics, and security scan status help verify the result; a successful commit alone does not establish a healthy demo.
05How can Hugging Face work with GitHub and Slack?+
Repository events can trigger a workflow that posts source-linked updates in Slack or creates GitHub issues for verified problems. Evaluation notes can live in Notion while the model or dataset revision remains the source of truth.
06Is finding a model the same as deploying an endpoint?+
No. SPIRITT can discover models, maintain repositories, generate chat completions, and create text embeddings. This connection lists inference endpoints but does not provision them, so using a model in an application is distinct from creating dedicated hosting.