Vlad Tansky

744 posts

Vlad Tansky banner
Vlad Tansky

Vlad Tansky

@vltansky

Israel Sumali Mayıs 2012
1K Sinusundan205 Mga Tagasunod
Guy Bary
Guy Bary@guy_bary·
Happy to share two new skills I’ve created to solve a couple of pain points I’ve been experiencing (with Octocode 🐙) 1. octocode-brainstorming Validates whether an idea is worth building before writing any code, using the Tavily API with Octocode search tools to ground decisions in real-world context (web and code). It researches GitHub, packages, and the web in parallel, then uses two AI agents to debate the same evidence. ** This skill requires 𝐓𝐀𝐕𝐈𝐋𝐘_𝐀𝐏𝐈_𝐊𝐄𝐘 (I'm using the free tier) 2. octocode-news Keeping up with tech news across multiple domains is fragmented, time-consuming, and often lacks depth. That’s why I created a skill that spawns sub-agents to conduct real research via RSS and curated sources, ultimately producing a full HTML report. skills.sh/bgauryy/octoco… skills.sh/bgauryy/octoco…
Guy Bary tweet media
English
2
4
7
465
Matt Pocock
Matt Pocock@mattpocockuk·
I might be renaming /grill-me to /discuss
English
72
1
322
27.1K
Vlad Tansky
Vlad Tansky@vltansky·
@AndrewGinns Nice! I have automation for it right now that check all my git activity + slack + calendar. This will level it up! Bring it to enterprise asap!
English
0
0
1
41
Andrew Ginns
Andrew Ginns@AndrewGinns·
Been using Codex Chronicle (Telepathy) since early in the dev cycle. One of the best use-cases I have for it is an automatic regular log of what I've done. We move so fast at OpenAI it's been a huge help to keep a log of what I've actually done each day. developers.openai.com/codex/memories…
English
2
0
57
2.7K
Kevin Kern
Kevin Kern@kevinkern·
codex app now includes chronicle (research preview). its an opt-in feature that uses recent screen activity to help codex understand what you're working on without you repeating as much context. memory stores what codex has already learned, while chronicle helps create better memories from your screen, so they work together. how to enable: 1. open settings in the codex app, 2. go to personalization, 3. make sure memories is on, 4. then turn on chronicle underneath it. can I use it? - needs GPT Pro - needs macos - NO EU,UK,Switzerland atm.
Kevin Kern tweet media
English
2
3
51
4.6K
Chrome
Chrome@googlechrome·
Heads up! Gemini in Chrome is coming to more Asian-Pacific countries! 🎉
English
118
224
2K
448.9K
Vlad Tansky
Vlad Tansky@vltansky·
@kr0der @mattymaddog_89 @rohanvarma You can setup tour own GitHub action that will run on every commit Didn’t mean you claimed 2x, I just said that I would not expect that high difference. IMO code review prompt + model + harness are the main factors
English
0
0
1
23
Anthony Kroeger
Anthony Kroeger@kr0der·
it's not 2x better (didn't claim that), but it's better than the standard codex review offering in my opinion in terms of how many bugs it catches and the fact that it runs on every commit whereas codex review doesn't (the limits aren't high enough if you do) if i had to set up a new project i'd just use bugbot + codex review bots (i've only tried 4 of them, the 4th being copilot 💀)
English
2
0
1
78
Anthony Kroeger
Anthony Kroeger@kr0der·
i think everyone should be subscribed to Cursor whether you use Cursor or not for coding, you need a sub to buy Bugbot. in my opinion Bugbot is just insanely valuable and it's worth buying a $20 sub just to access it. it's $40/m for unlimited PR reviews and the accuracy is insane, you can tell that it digs deep into the code rather than just taking the PR at face value. i haven't even bothered looking for a better AI PR review tool because it's too good already. definitely my most valuable non-codegen AI subscription by far
English
25
11
233
19.8K
Vlad Tansky
Vlad Tansky@vltansky·
@kr0der @mattymaddog_89 They are not training bot, they are building product around it. I agree that its better, but I’d be curious to see a real benchmark comparison - i would not expect huge difference - not 2x better imo @rohanvarma do you happen to have a comparison?
English
1
0
2
48
Anthony Kroeger
Anthony Kroeger@kr0der·
@mattymaddog_89 there's no chance me running a custom review prompt to claude code/codex is gonna beat a whole funded team training an AI review bot...
English
2
0
2
188
Peter Steinberger 🦞
Peter Steinberger 🦞@steipete·
it took a hammer, but i managed to beat emotions into GPT.
Y@agi_ytml

@steipete Ok works well for me, getting the old claude vibes, thanks a lot 🙌🏽

English
97
40
1.4K
240.3K
Vlad Tansky
Vlad Tansky@vltansky·
Used dev-browser vs agent-browser to find the best sushi combo on Wolt. Feels like a pretty good use case. Ran both twice in subagents with the same starting context. Dev-browser was 3x faster with 33% less tokens used!
Vlad Tansky tweet media
English
1
0
4
1K
Ryan Carson
Ryan Carson@ryancarson·
Ryan Carson tweet media
Andrej Karpathy@karpathy

Wow, this tweet went very viral! I wanted share a possibly slightly improved version of the tweet in an "idea file". The idea of the idea file is that in this era of LLM agents, there is less of a point/need of sharing the specific code/app, you just share the idea, then the other person's agent customizes & builds it for your specific needs. So here's the idea in a gist format: gist.github.com/karpathy/442a6… You can give this to your agent and it can build you your own LLM wiki and guide you on how to use it etc. It's intentionally kept a little bit abstract/vague because there are so many directions to take this in. And ofc, people can adjust the idea or contribute their own in the Discussion which is cool.

ZXX
7
1
45
17.6K
andrej
andrej@reactive_dude·
OK so apparently this other CRITICAL issue GPT found shud be downgraded to MINOR. I should just connect them to a chat so they can talk it out.
andrej tweet media
English
10
0
110
30.2K
andrej
andrej@reactive_dude·
Opus 4.6: all good, ship the PR GPT 5.4: two critical issues, one major issue 😬😬😬
English
95
44
2.9K
364.2K
andrej
andrej@reactive_dude·
What are your favorite agent skills? I'll start: > grill-me (brainstorming) > write-a-prd (specs) > tdd (the best way to code with agents rn) > agent-browser (great for debugging/qa)
English
35
22
715
70.1K
Vlad Tansky
Vlad Tansky@vltansky·
@mattpocockuk I treat plan as a helper tool for llm. To keep it on track. Plan stored as file + generate todo list which help to keep model on track on long running missions + compactions
English
0
0
0
69
Matt Pocock
Matt Pocock@mattpocockuk·
I have also stopped using plan mode It creates a plan FAR too eagerly and usually asks you zero questions en route The whole point of planning is to get on the same wavelength with the LLM, not to generate an asset you don't read /grill-me all the way
Peter Steinberger 🦞@steipete

I never use plan mode. The main reason this was added to codex is for claude-pilled people who struggle with changing their habits. just talk with your agent.

English
171
48
1.4K
283.6K
Matt Pocock
Matt Pocock@mattpocockuk·
Today's fun job: building an AI SDK-style API for running different agents in containers. Soon Codex will be able to review CC's work, programmatically. Sandcastle is looking nice.
Matt Pocock tweet media
English
31
12
295
19.7K
Ryan Carson
Ryan Carson@ryancarson·
Here's what my openclaw does for me: 1. Schedules all my meetings and manages my calendar 2. Daily finds new business prospects, updates a Google Sheet, and reaches out to setup meetings 3. Prepares my task list for the day before I wake up and then manages it throughout the day. 4. Regularly follows up on emails that haven't gotten a response. 5. Backs himself up to GitHub every night. I've been considering the idea of setting up my claw on X and having him run his own little business where he helps people set up their openclaw with the cron jobs and skills he uses to do all this. Would anybody pay for that? If you would pay, what's it worth to you? I'm thinking it's a downloadable package of all the JSON and Markdown that's required to get this set up + some instructions, and then he would provide email support.
English
56
5
149
13.3K
Vlad Tansky
Vlad Tansky@vltansky·
Missing ⁠ /simplify ⁠ from Claude Code in Codex? One click install: `npx skills add vltansky/skills/skills/simplifier -g -y`
Anthony Kroeger@kr0der

i got Claude Code to tell me what its /simplify prompt is so i could copy paste it into Codex, since Codex loves writing unnecessary and duplicate code. bookmark this and add it to your Codex prompts: --- description: "Review changed code for reuse, quality, and efficiency, then fix any issues found." --- # Simplify: Code Review and Cleanup Review all changed files for reuse, quality, and efficiency. Fix any issues found. ## Phase 1: Identify Changes Run `git diff` (or `git diff HEAD` if there are staged changes) to see what changed. If there are no git changes, review the most recently modified files that the user mentioned or that you edited earlier in this conversation. ## Phase 2: Launch Three Review Agents in Parallel Launch all three agents concurrently in a single message. Pass each agent the full diff so it has the complete context. ### Agent 1: Code Reuse Review For each change: 1. **Search for existing utilities and helpers** that could replace newly written code. Look for similar patterns elsewhere in the codebase — common locations are utility directories, shared modules, and files adjacent to the changed ones. 2. **Flag any new function that duplicates existing functionality.** Suggest the existing function to use instead. 3. **Flag any inline logic that could use an existing utility** — hand-rolled string manipulation, manual path handling, custom environment checks, ad-hoc type guards, and similar patterns are common candidates. ### Agent 2: Code Quality Review Review the same changes for hacky patterns: 1. **Redundant state**: state that duplicates existing state, cached values that could be derived, observers/effects that could be direct calls 2. **Parameter sprawl**: adding new parameters to a function instead of generalizing or restructuring existing ones 3. **Copy-paste with slight variation**: near-duplicate code blocks that should be unified with a shared abstraction 4. **Leaky abstractions**: exposing internal details that should be encapsulated, or breaking existing abstraction boundaries 5. **Stringly-typed code**: using raw strings where constants, enums (string unions), or branded types already exist in the codebase 6. **Unnecessary JSX nesting**: wrapper Boxes/elements that add no layout value — check if inner component props (flexShrink, alignItems, etc.) already provide the needed behavior 7. **Unnecessary comments**: comments explaining WHAT the code does (well-named identifiers already do that), narrating the change, or referencing the task/caller — delete; keep only non-obvious WHY (hidden constraints, subtle invariants, workarounds) ### Agent 3: Efficiency Review Review the same changes for efficiency: 1. **Unnecessary work**: redundant computations, repeated file reads, duplicate network/API calls, N+1 patterns 2. **Missed concurrency**: independent operations run sequentially when they could run in parallel 3. **Hot-path bloat**: new blocking work added to startup or per-request/per-render hot paths 4. **Recurring no-op updates**: state/store updates inside polling loops, intervals, or event handlers that fire unconditionally — add a change-detection guard so downstream consumers aren't notified when nothing changed. Also: if a wrapper function takes an updater/reducer callback, verify it honors same-reference returns (or whatever the "no change" signal is) — otherwise callers' early-return no-ops are silently defeated 5. **Unnecessary existence checks**: pre-checking file/resource existence before operating (TOCTOU anti-pattern) — operate directly and handle the error 6. **Memory**: unbounded data structures, missing cleanup, event listener leaks 7. **Overly broad operations**: reading entire files when only a portion is needed, loading all items when filtering for one ## Phase 3: Fix Issues Wait for all three agents to complete. Aggregate their findings and fix each issue directly. If a finding is a false positive or not worth addressing, note it and move on — do not argue with the finding, just skip it. When done, briefly summarize what was fixed (or confirm the code was already clean).

English
1
0
4
184