Repository navigation
Decide: enable Wikipedia/GBIF enrichment in prod or remove the External Data page #461
Description
Decision plan — 2026-10-15
Delivery plan: #558. This is the existing decision ticket, not a second enrichment implementation project.
- Verify current production provider configuration and the admin status using read-only evidence. The original report below describes disabled providers; this planning review confirmed the disabled-client code path but did not independently inspect production config.
- Compare two operating models: supported Wikipedia/GBIF sync with a named operational owner, versus removing the admin page/routes while retaining intentional manual tooling.
- If enabling: define timeouts/retries, attribution/license preservation, honest disabled/running/failed states, observed successful manual sync, schedule, monitoring and rollback. Choose explicitly whether URL-only images suffice; R2 copying is additional scope.
- If removing: identify navigation/routes and tests to remove, document retained manual tooling, and resolve Deploy External Data Cron to Production #249 as superseded if scheduling is no longer desired.
- Record the maintainer's chosen model and rationale, then scope implementation only for that branch. Use the current Fly runtime; do not follow the obsolete EC2 cron instructions.
Blocked by: None. This decision can proceed alongside report and collection fixes; production enablement is not authorized merely by filing this plan.
Original problem report
/admin/external-data always shows "Last run failed" in prod. The prod config sets wikipedia.enableSync and gbif.enableSync to false, so the clients throw "disabled in configuration" (src/integrations/wikipedia.ts:392, gbif.ts:307).
Choose one:
- Enable and schedule it (Deploy External Data Cron to Production #249). Also fix attribution being dropped and the SPARQL call bypassing the base client's timeout and retry (
wikipedia.ts:210). - Remove the admin page and routes and keep the scripts for manual runs.
Activity
Remove the in-app External Data page; enrichment runs off-box (#461) (#…
Push external species data to prod over the admin API (#595, #461)
Push enriched Species data to prod over the admin API, so enrichment …
jra3 commentedon Oct 28, 2026
Note
🤖 Claude Opus 5.5 responding on behalf of John Allen
Decided: enrichment runs off-box. #596 removed the in-app page and sync; #601 added
scripts/push-external-data.ts, which pushes to prod through the admin API. The pipeline is indocs/EXTERNAL_DATA_CRON.md.