Deploy from the Browser

Use this page to create a private FitOps dashboard on HuggingFace Spaces without installing Python or running the CLI locally.

The browser wizard sends your deployment details to the FitOps Deploy API. The API runs the same HuggingFace deployment flow as fitops deploy hf, using the HuggingFace and GitHub tokens you provide for this one deployment job.

Security model

Tokens are sent to the Deploy API for the active job only. The hosted API keeps them in memory, redacts them from job events, and drops them when the job finishes or expires. If you prefer not to send tokens to the hosted API, use the local CLI deploy flow.

What You Need

Requirement	Why
HuggingFace account	Owns the private Space that runs your dashboard
HuggingFace write token	Lets FitOps create/update your Space
GitHub account	Owns the private backup repository
GitHub token with repo scope	Lets FitOps read/write backup releases and configure keepalive
Authenticator app	Stores the dashboard two-factor login key
Strava API app	Optional during deploy, but needed before activity sync

Strava Client ID and Client Secret are optional in the wizard. If you provide them, FitOps stores them in the deployed Space and sends you straight to Strava OAuth on first setup. OAuth is still required before FitOps can sync activities.

Deploy Wizard

The docs site is not configured with VITE_FITOPS_DEPLOY_API_BASE_URL, so the live wizard is disabled on this build.

Deploy FitOps Online

Create a private HuggingFace Space, protect it with password + 2FA, and optionally prefill Strava setup so your first dashboard login goes straight to OAuth and sync.

HF tokenGitHub tokenStrava API settingsStrava guideFitOps logo

HuggingFace tokenHF Space repoOptional. If empty, FitOps uses your HF username and fitops-dashboard.GitHub tokenGitHub backup repoStrava Client IDOptional. Find it on the Strava API settings page.Strava Client SecretOptional. This pre-fills setup; OAuth still happens after deploy.Dashboard passwordConfirm password

If you leave the HF Space repo empty, the exact Strava callback domain appears after deploy.

After Deployment

1. Add the generated Authorization Callback Domain in Strava API settings.
2. Open the dashboard URL shown by the wizard.
3. Sign in with your dashboard password and authenticator code.
4. Complete Strava OAuth in the deployed dashboard.
5. Wait for the first sync to finish, then open the dashboard.

Prefer the Terminal?

The local CLI flow remains available and does not call the hosted Deploy API:

fitops deploy hf \
  --hf-token hf_xxxxxxxxxxxxxxxxxxxx \
  --hf-repo yourname/fitops-dashboard \
  --github-token ghp_xxxxxxxxxxxxxxxxxxxx \
  --github-repo yourname/fitops-backups \
  --strava-client-id 175267 \
  --strava-client-secret xxxxxxxxxxxxxxxxxxxx

See HuggingFace Spaces for the full local deployment reference.

Deploy API Operator Notes

The browser wizard expects a hosted API running the FitOps deploy app. This repo includes a HuggingFace Space Docker context at fitops/cloud/deploy_api_space/ and a GitHub Actions workflow that pushes it.

Create these GitHub repository settings before running Deploy FitOps Deploy API Space:

GitHub setting	Type	Purpose
HF_TOKEN	Secret	HuggingFace token allowed to create/update the Deploy API Space
FITOPS_DEPLOY_API_BASE_URL	Variable	Public Deploy API base URL injected into the docs build
FITOPS_DEPLOY_API_HF_SPACE_REPO	Variable	Target Space repo, for example yourname/fitops-deploy-api
FITOPS_DEPLOY_API_SPACE_URL	Variable	Public Space URL used by the keepalive workflow
FITOPS_DEPLOY_ALLOWED_ORIGINS	Variable	Comma-separated docs origins allowed to call the API

Optional GitHub variables:

Variable	Default	Purpose
FITOPS_DEPLOY_RATE_LIMIT_PER_HOUR	5	Job creation limit per client IP
FITOPS_DEPLOY_MAX_CONCURRENT_JOBS	2	Maximum in-process deploy jobs
FITOPS_DEPLOY_JOB_TIMEOUT_SECONDS	900	Deploy timeout
FITOPS_DEPLOY_JOB_TTL_SECONDS	3600	In-memory job retention

The workflow creates or updates a private Docker Space and stores the runtime configuration as Space secrets. A second workflow pings /health every 20 minutes so the Deploy API Space stays warm.

For a manual API runtime, run:

uvicorn fitops.cloud.deploy_api:app --host 0.0.0.0 --port 7860

Configure these environment variables on the API runtime:

Variable	Required	Purpose
FITOPS_DEPLOY_ALLOWED_ORIGINS	yes	Comma-separated docs origins allowed to call the API
FITOPS_DEPLOY_RATE_LIMIT_PER_HOUR	no	Job creation limit per client IP, default 5
FITOPS_DEPLOY_MAX_CONCURRENT_JOBS	no	Maximum in-process deploy jobs, default 2
FITOPS_DEPLOY_JOB_TIMEOUT_SECONDS	no	Deploy timeout, default 900
FITOPS_DEPLOY_JOB_TTL_SECONDS	no	In-memory job retention, default 3600

Configure the docs build with the public API base URL:

VITE_FITOPS_DEPLOY_API_BASE_URL=https://your-deploy-api.example.com npm run docs:build

The API stores jobs in memory only. Restarting the API clears running and completed jobs.