Retry Desk

Read a retry policy written one hop at a time and see what the whole chain costs. A hop's worst case is attempts x timeout + sum(backoff), not its timeout: 3 x 10s with 1s doubling is 33s, already more than a 30s client deadline on its own, and 1.65m three hops deep. Attempts multiply along a chain rather than adding, so three hops at three attempts is 27 leaf requests for one client call - nobody configured 27, each team configured 3. The deadline is spent in chain order, so unless it propagates the innermost hop keeps working after the caller has given up: 7 of the 9 attempts in an ordinary three-hop chain begin after the client has already gone. Without jitter the whole population retries at the same instant, at an offset the backoff makes exactly predictable. A 20-request window at 50% trips a breaker on 10 failures and one half-open probe re-closes it wrongly 50% of the time. Five lanes over one chain sheet: plan a policy the deadline can pay for, read what an existing chain costs, price the budget and the load, and decide what to change. Free browser-side arithmetic - no request is made to any hop and nothing is measured.

Every public app is built from a security-scanned skill and must pass a clean scan — skill and frontend — before it can be listed. Have a skill of your own? Turn it into an app — or read the step-by-step walkthrough.