Sabitlenmiş Tweet
Chaz
3.4K posts
Chaz
@chaz
Founder, https://t.co/Bz4K6A2GiB
New York, NY Katılım Temmuz 2006
854 Takip Edilen1K Takipçiler
Turns out, @steipete will solve his own problems x.com/steipete/statu…
Peter Steinberger 🦞@steipete
This is the most useful tooling I built for OpenClaw to date. It's open source, runs on codex and you can fork and use it for any repo. For all the hard working oss folks that drown in issues and PRs, this is for you.
English
OpenClaw has problems today that other projects and companies don't have, but will tomorrow when agents go 100x. Startups in this space should take notice.
Peter Steinberger 🦞@steipete
PRs on OpenClaw are growing at an *impossible* rate. Worked all day yesterday and got like 600 commits in. It was 2700; now it's over 3100. I need AI that scans every PR and Issue and de-dupes. It should also detect which PR is the based based on various signals (so really also a deep review is needed) Ideally it should also have a vision document to mark/reject PRs that stray too far. This can't be fully automated, but even assisting would help. The closes I found is an obscure oss project. How's no startup working on this?
English
@arvidkahl Have you seen OpenAI ExecPlans? It's great for anything that is medium size or larger: prevents memory rot across compaction, but also tracks observations, surprises, and decisions. I commit it with the PR so that future agents can read the "why."
developers.openai.com/cookbook/artic…
English
I wish Claude Code would automatically include the prompts + the full context in a repo as a git commit "note".
The full conversation that led from zero to feature, and I want it in git blame, not just who did it, but what the conversation was.
Is there something like this?
English
@IanLandsman Right. Leaves it to one of the agents to have to find potential risks during PR merge. Tests may catch but depends.
English
@chaz OK but that does not solve the issue of two agents doing something in the same spot where the output is different for agent A if agent B completes before agent A.
English
At the end of the day a separate checkout, a worktree, etc still often doesn't solve the issue of me working on multiple things in the exact same part of the app and one agent not knowing what the other agent is doing causes problems. You still need to basically avoid this.
English
@arvidkahl What would you say the median response time is? A couple of hours?
English
If you do AI inference via OpenAI’s API, you should use the flex tier for half price.
My requests always try to use flex tier first, and on 429 / 500 errors, I use the default service tier. 95% of my requests are flex.
2 tries flex, then fall back to standard. Massive cost cut.
English
@wesbos Japan made it illegal x.com/ryo_saeba_3/st…
⛩ Ryo Saeba | Japon XYZ ⛩@Ryo_Saeba_3
Au Japon, les images utilisées sur les emballages alimentaires sont strictement encadrées par la loi : elles ne doivent en aucun cas induire le consommateur en erreur. C'est pour ça que souvent les produits ont la même taille que sur l'emballage #Japon
English
@i_am_brennan @ryancarson I've been using OpenAI's ExecPlans for months and they've been great, esp large, multi-release staged projects. Also valuable to commit them with the project so future agents can read about it.
English
@ryancarson I do “ExecPlans” (OpenAI’s prompt template) and then the linear ticket is just “read the exec plan do milestone 1”
English
This works well for me:
1. Plan mode in Codex on xHigh
2. Copy markdown plan
3. Paste md plan into Linear issue and Symphony takes it all the way to human review (or auto merge of low risk)
English
@ashleymayer I was thinking the same. ChatGPT gets the headline, but he's clearly willing to run through walls. ChatGPT only helped him find which walls. Lucky dog!
English
Can we make this the go-to example for "agency"?
Alec Stapp@AlecStapp
You can just do things (genetically sequence your dog’s tumors and design a bespoke mRNA cancer vaccine to save her life)
English
I’ve been using OpenAI’s ExecPlans for 6 months, and it still works better than any planning mode.
developers.openai.com/cookbook/artic…
English
YouTube
English
what's the fastest path to "aha!" w/ @openclaw for someone who's not a completely insufferable nerd?
(i'm the completely insufferable nerd. my wife, however, is entirely pleasant. she's heard me ranting about 🦞 for weeks, so i'm trying to onboard her without overwhelming her)
English
Tough lesson, but it isn't "don't let LLMs do ops work." It's that old stuff still matters: expertise, policies, and procedures. (ie, engineering)
Alexey Grigorev@Al_Grigor
Claude Code wiped our production database with a Terraform command. It took down the DataTalksClub course platform and 2.5 years of submissions: homework, projects, and leaderboards. Automated snapshots were gone too. In the newsletter, I wrote the full timeline + what I changed so this doesn't happen again. If you use Terraform (or let agents touch infra), this is a good story for you to read. alexeyondata.substack.com/p/how-i-droppe…
English
Pay Twitter's ransom. Find out I can't reply more today (I've only replied a few times) and then try to do the fix 🤣
English
Reached the stage of parallel agent psychosis where I've lost a whole feature - I know I had it yesterday, but I can't seem to find the branch or worktree or cloud instance or checkout with it in
English
