Tokenflux
Field guide
Notes we find ourselves sharing with clients. Use them whether or not you work with us.

Webhooks that do not lose events
Acknowledge quickly and process later: return a success response as soon as you have stored the payload, then do the work from a queue. Verify signatures on every request. Assume events can arrive twice and out of order, and design handlers to tolerate both. Finally, schedule a reconciliation that compares both systems periodically; webhooks are a speed optimisation, not a guarantee.
Living with rate limits
Read the provider's limit headers rather than guessing. Use exponential backoff with some randomness so retries do not stampede. Batch where the API allows it. Size bulk jobs to fit comfortably inside quotas, and alert when queues grow faster than they drain.
Surviving API version changes
Pin the API version you build against if the provider allows it. Subscribe to the changelog and deprecation notices with a shared mailbox rather than one person's inbox. Record upgrade deadlines next to credential expiry dates.
Credentials without anxiety
Use service accounts, not personal logins. Store secrets in your hosting platform's secret manager, never in source code or chat. Record who can rotate each credential and when it expires. Never email passwords or keys — to us or anyone.
Data mapping is a business conversation
“Customer” rarely means the same thing in two systems. Map fields with the people who use them, write down what happens when values are missing, and agree which system owns each field.
Personal data
If personal data moves between services, your data protection obligations travel with it. Move the minimum fields needed, and check your agreements with each provider. We can describe data flows clearly for your records; we do not give legal advice.