Haven't seen codex writing ad-hoc codemods before, but it just did for a bigger TypeScript migration. Impressed.
@steipete
-
Autoreview keeps GPT honest and helps achieve real goals
By
–
gotta use autoreview, that keeps gpt honest and usually helps achieve the real goal.
-
Woke up to codex finishing SMS task and your tweet on shared idea
By
–
Did we had the same idea yesterday? Didn’t see your proposal, to wake up with my codex finishing the sms channel task and your tweet.
-
Use Codex instead of ChatGPT for better responses
By
–
You need to ask codex, not chatgpt if you want good responses.
-
GPT 5.5 boosts prompt duration and confidence with new features
By
–
With GPT 5.5, /goal, autoreview and crabbox my prompts moved from ~30-60min to often 4-10h tasks and my confidence that it’s ready is much much higher. Yielding agents is a skill.
-
Agents assume wrong invariants; fix via .md constraints explanation
By
–
Most of the time. Sometimes the agent assume the wrong invariants, and the real fix is text in agents .md to explain system constraints.
-
Spinning up a crabbox with webvnc for a test environment
By
–
I ask it to spin up a crabbox with webvnc for a fully prepped test env so I can click through it myself or as Mantis for videos.
-
Codex review: all good until told there is a bug
By
–
I do this with codex all the time. Ask it to review code for bugs and it will tell you all good, tell it there is a bug and it will LOOP AND LOOP and will find issues.
-
Shift from Opus to GPT 5.5 as best model for Clawdbot
By
–
I was full on promoting Opus as best model for “Clawdbot”. Luckily that changed and GPT 5.5 is now the best model based on our internal benchmarks.
