Two bugs from my own n8n product, both of which passed on my machine:
- require(‘crypto’) — sandboxed away on a stock install. Worked locally, dead on a clean instance.
-
- an $env read that throws under N8N_BLOCK_ENV_ACCESS_IN_NODE, which had quietly turned a fail-safe gate into a fail-crash gate.
- Neither showed up in development. Both would have shipped. They were caught by acceptance tests I wrote before building, and that’s the whole of what I sell: I write down what “fixed” means before I touch anything, you agree it, and production writes and sends stay disabled until it passes. You get evidence against each criterion and a note on what actually caused it — not “fixed it”.
Same-day fixes I’ll take, one at a time:
- a workflow that fails intermittently and nobody can say why
-
- a webhook that returns 200 and does nothing, or fires twice
-
- duplicates appearing, or a dedupe that quietly stopped deduping
-
- an integration broken after a provider changed something under you
-
- transactional or outbound mail landing in spam — SPF/DKIM/DMARC, alignment, list hygiene
- $120 for one issue, up to two hours. If it’s bigger than that I tell you before I start and quote it separately rather than running the clock. Half up front, half when the criteria pass. Wise or PayPal.
- Straight with you: I work solo out of Poland and I don’t have a client portfolio yet. No agency overhead is why the number is what it is, and it’s why I’d rather win the first few clients on price than on promises. If that’s disqualifying, no hard feelings.
- First thing I’ll ask you: what broke last time, and how did you find out? That usually tells me more than the logs do.
- Reply here or email me and I’ll answer today.
— Kacper Konieczny, niekonieczny.k@gmail.com