opencode/services/data

Data

Deploys the opencode.ai/data site from packages/data as three Workers with cf:

Worker	Config	What it is
opencode-data-<stage>	cloudflare.config.ts	The SolidStart site, on /data* and /<locale>/data* of the stage's host
opencode-data-og-<stage>	og.config.ts	Open Graph images, reached only through the site's OG_WORKER binding
opencode-data-sync-<stage>	sync.config.ts	The opencode-data-sync-<stage> Workflow and the cron handler that starts it

script/build.ts builds all three into one Build Output tree and script/deploy.ts deploys them with cf deploy --prebuilt, the OG Worker first. The data job in .github/workflows/deploy.yml runs deploy:dev or deploy:production and passes the Worker secrets from AWS Secrets Manager (opencode/<stage>/DATA_DATABASE_HOST, DATA_DATABASE_USERNAME, DATA_DATABASE_PASSWORD, R2_SQL_AUTH_TOKEN, DATA_HIDDEN_MODELS) with --secrets-file. Each Worker gets only the secrets its config declares.

The database is the opencode-stats PlanetScale database: the dev branch for dev and production for production, through the data-worker-<stage> password. It is not managed as code.

Every config, script/build.ts and script/deploy.ts take the stage from --mode, which must be dev or production (stage.ts). Anything else, including no mode, throws instead of falling back to production, so pass --mode to cf commands that read the configs too (cf deploy --dry-run --mode dev).

bun run dev               # the site on your machine (packages/data/app)
bun run build:dev         # or build:production
bun run typecheck