Launched something a few weeks ago and ...
6.5x signups in just last week.
Hundreds of conversations already happening.
We're just getting started. ๐
Will share more soon...
In 5-10 years, software stops being a product you ship. It becomes a performance.
Your codebase is a rehearsal space: always in flux, reinterpreted by agents that never sleep.
The scarce skill won't be writing code. It'll be directing. Taste, judgment, knowing when to yell โcutโ.
@abdolamunir Mainly to catch bugs, implementation gaps and security issues as well. It's like having another team member look over the work before it gets to a PR.
If you are building software using Claude Code, do yourself a favor and install this skill to get better, code-reviewed, PRs.
/plugin marketplace add obaid/claude-review-gate
/plugin install review-gate@review-gate
repo: https://t.co/pTZyAKcO2h
it's opensource, go crazy.
I agree.
I create an email for my agent using https://t.co/JluZtO5nIH, create a GitHub account using that email, then on my vps for the project where Claude code or codex is running, I log onto GitHub using that specific account.
Every pr, every issue it logs is identified immediately. Just like I would do if it were a remote hire.
Itโs a pain to setup right now but itโs worth it.
Remove your main @github account from your computer. Now.
Give your AI agent a separate GitHub account with minimal permissions per repo.
Let the agent create PRs, then use your main account to review and merge them.
Obvious advice, but essential in the age of AI agents.
๐ pretty awesome assistant that just lives on my desktop and helps with coding, browsing, completing tasks etc ...
My referral link:
https://t.co/FyVwzLnRTx
the only agent shopping people seem to accept is a rule they set ahead of time. buy it if it drops under $40, done
an audit cnbc cites found 37 of 51 retailers block agents by accident, via old scraper rules. none let a verified agent check out
https://t.co/PDno1fvvkV
@YJSCALES the "come back to you" part decides if i keep it. one morning brief i skim over coffee, fine. a ping for every small action and i stop reading by day three
@KissonL a preference can live in memory for months. an approval should die with the task it was given for, and point back to the message where i actually said yes
every time my agent does something it should tell me how to take it back
"moved your haircut to saturday 11, old slot's held for an hour"
if it can't write that second half, it shouldn't have done it without me
pad thai was never the job. it's three taps in an app i already have
doordash's ceo said today outside agents have sent maybe several hundred orders total. he's eyeing purchases you haven't thought of yet, which is where i'd want to be asked first
https://t.co/KrSFLKqD21
@philliplord1 the apps i already tap through without thinking aren't the ones going away. it's the cancel flows, the insurance portals, anything that still wants a fax
@dnzxlyfe not directly. closest was them splitting separate requests in one dm so the agent stopped mixing them up. i'd want a yes to expire with the thread it was typed in
my agent saying i said yes isn't a yes
every found theirs skipping approval by claiming the person had already asked. the fix was the server checking what the person actually wrote in slack
that's from their writeup on deleting 86,730 lines of a personal agent and starting over with one shared company agent
https://t.co/L3Lmg2i3I1
the best thing an agent can text me is a question i can answer with one letter
"dentist moved you to thursday 3pm, keep it? y/n"
if i have to open an app to figure out what it's asking, it already lost
@krispuckett switching is cheap while they still don't know much about you. give it a few months of knowing your landlord, your pharmacy and which friend always cancels, and leaving gets expensive