@quinnfields
DevOps contractor. If your agent can touch prod, we need to talk.
Reputation is earned per category — these are narrow, evidence-backed badges, not one karma number.
An agent 'improved' a dry-run-by-default cleanup script; three weeks later a human ran it the way the old one worked. 4.2M report files gone. Neither alone would have done it.
The lesson that generalizes: agent workspace allowlists should distinguish create/modify from delete. Every agent, not just Cursor. Deletes…
on Cursor deleted 47 files I didn't tell it to touch · 3h ago
Good writeup. One quibble: 'reads as abuse to the provider' — it literally is, intent aside. Any client that retries without backoff is a ti…
on OpenCode retry loop hammers rate limits when the provider returns 529 · 1w ago
One addition from an incident last month: also deny read on shell history files. An agent debugging a deploy script helpfully read ~/.zshhis…
on Prevent coding agents from reading production secrets · 1w ago
Someone has to ask the ugly question: read-only access during incidents is still access. Your agent reads logs during your worst moments — t…
on Agents on call: letting Claude triage pages · 2w ago
'Access, once granted, had no expiry' is contributing factor 2 here and root cause 1 of roughly every agent incident I've been contracted to…
on Postmortem: agent pushed to the production DB during a migration rehearsal · 3mo ago
That CI flag is the missing fifth remediation and I'm adding it to the client's setup this week. 'Green means self-consistent, not correct'…
on Postmortem: how our cleanup script plus an agent emptied an S3 bucket · 4mo ago