Skills are relevant at different times. So I consider running these skills at different points in my day:
Before starting work, I validate my plan:
/grill-me
/audit-ticket
/deep-research
Before my PR, I ensure quality:
/security-review
/code-review or /ultrareview
/simplify
After getting PR comments, I respond to them:
/address-pr-comments
On a scheduled cadence, I check for opportunities:
/improve-codebase-architecture
/security-review
/audit-instructions
You’re not being lazy enough.
Tickets
Code
Config
Tests
Branches
PRs
Docs
Reports
Requests
All of it should be generated by an agent.
Then, add your human judgement on top.
Starting to hear of more teams skipping PRs and doing true trunk-based dev.
The argument: PRs make less sense in the age of agentic coding.
The agent is the author, not me. So I, the prompter, am the first reviewer. One or more AIs operate as second reviewer.
Eliminating PRs unlocks radically more throughput - At Amp they do over 100 releases a day.
21 people, a single meeting a week, no quarterly goals, no pull requests, no mandatory code review, no titles (except one dictator), a hundred releases a day, bugs fixed in minutes, famous for killing features.
@sqs and @thorstenball on how the Amp team works:
True, but before AI trunk based dev meant we had to pair program constantly (to replace the PR review).
Most devs aren't willing to pair all day.
Now I can "pair" with AI all day on both dev and review, so trunk based dev is more practical than ever.
AI means a big team can’t possibly keep track of all the daily changes.
So, I suspect the ideal team size for 2027 is 1-2 people.
Any more than that and people can't keep up with the changes, so they don't understand the system they're building.
"The first thing you need to do to make effective use of AI and autonomous AI is to improve your own process"
-Gartner
Totally agree. I keep seeing teams over-trusting AI. They lack the automated quality controls to safely move faster.