# Workflow discover - use sourcegraph to get new graphql urls - for all entries with no status or status failed: - try fetching their schema - if no title, description, website, add them to entry with website ones - try generating npm package with no publish - mark them as success or failed - add website, title, description adding to website - manually mark apis in csv as enabled, optionally add example query hints - for each enabled package with no previous published version (also add npm scope so if i change it it will run again) - generate npm package - generate example queries code - publish package (if not dry) - save generated queries, favicon, version to generated.csv - if not dry: add package to csv with @scope/slug and version with 0.0.0 updating npm package (once a day?) - clean up error message from csv - if schema changed increase minor version - generate again package - generate again queries, favicon - if generation failed, add error message to csv - publish with new version - generate changelog? later maybe so manual steps are: - discover new urls with `pnpm discover` and add them to csv - add slug to apis that i think are interesting - try generating package with `pnpm publish:dry`, inspect package folder, see generated queries and such # considerations where should i put generated code? if it's same csv file it may become difficult to edit it