Browse docs ▾
Use cases
"My automation broke but I only changed the inputs"
When code didn't change but a run started failing, work in this order — authentication first, then connectivity:
1. Auth first. Open HTTP ▸ Auth check, pick the platform and run it. "Reached, but refused" (401/403) means the credentials/token are the problem — expired, rotated, or wrong for the new inputs. "Reached and authenticated" means auth is fine; move on. "Not reached" sends you straight to step 2.
2. Then connectivity. If auth is good but calls still fail, confirm you can actually reach the host: Ping and Path ping the API's hostname, and check IPv4 vs IPv6 (the new input may resolve to an address family that's filtered).
3. Re-run only after both pass. If auth succeeds and connectivity is clean, the failure is almost certainly in the inputs/data itself — validate the changed values.
Tip: a rotated key or certificate is a classic culprit — check the SSL inspector's "days remaining" and whether a token was rotated out from under the job.