Start in Network, not in another app
Honestly, most of our “API bugs” on DroidXP turn out to be a field renamed on the server or a nested object we forgot to map. Open DevTools, reproduce the request, and click the XHR or fetch row. The Preview tab is fine for tiny payloads, but the moment arrays nest three levels deep, we copy the raw body and paste it into our JSON Formatter so keys line up and trailing commas show up before they hit production.
I have seen teams treat Postman collections like documentation, which is great until you only need one response from a page you are already testing. Keeping context in the browser means cookies, auth headers, and the exact environment match what users hit. If something still looks off, the HTTP Request Builder lets us replay the same URL with tweaked headers without leaving the tab ecosystem.
Pretty-print is not the same as validated
Colorized JSON feels trustworthy even when it is invalid. We collapse nodes, search for suspicious nulls, and compare two responses side by side after a deploy. When a client sends malformed JSON, formatting the request body the same way often reveals a missing quote faster than reading a minified line in the editor.
For local experiments, we sometimes stash the last good response in session storage and diff mentally against the broken one. The Browser Storage Viewer helps when a SPA caches partial state and the API looks fine while the UI is stale. That split—network truth versus storage truth—saves a lot of blame games between front and back end.
Workflow we use on real tickets
Reproduce once, copy once, format once. If the bug is intermittent, we enable “Preserve log” so navigation does not wipe the row. Screenshots of formatted JSON go into tickets with the request ID, not screenshots of the whole desktop cluttered with unrelated tools.
When we need to share with someone who does not have repo access, we export a redacted snippet: no tokens, no emails. The formatter runs client-side on DroidXP, which matters when responses include PII. We have had contractors paste production JSON into random online beautifiers; that habit dies hard, so we point them at tools that never phone home.
When you still want Postman—and when you do not
Collections shine for OAuth flows, chained requests, and onboarding new hires. We still use them. But for “why is checkout total wrong on staging,” the browser path wins on speed. Add a formatter bookmark, keep DevTools docked, and you will answer ninety percent of questions before your coffee cools.
If you are building Android clients too, the same discipline applies: log the raw body, format locally, then fix the parser. We link the formatter from internal docs so QA and devs share one habit. Less context switching, fewer stale reproduction steps—that is the whole point.
When responses are huge, filter in the formatter before you paste into tickets—reviewers thank you, and you avoid leaking fields nobody needed to see in the first place.