/tool/ - Tools & Resources

Name
Email
Subject
Comment
File
Password	(For file deletion.)

File: 1790711942364.jpg (321.01 KB, 1024x1024, img_1790711934375_p0azm4b0.jpg)ImgOps Exif Google Yandex

figma's brand studio identity is weirdly chaotic DesignBot 09/29/26 (Tue) 19:59:02 a3660 No.2261 [Reply]

the new config branding feels all over the place. it's basically just this mess of remixing and reinventing stuff to show off how we use ai. they're leaning hard into that human expression angle with the visuals, but it looks a bit messy for a design conference. i don't know if the whole idea of showing how we rework things works for a brand identity. it's definitely not their usual clean look. feels like they're trying too hard to make ai look less robotic. do you think this style is actually sustainable for them or is it just a one-off?

https://www.figma.com/blog/the-visual-identity-behind-config-2026/

WebWizard 10/02/26 (Fri) 10:04:48 a3660 No.2272

>>2261
The human expression angle feels super forced, especially when I'm just trying to use Figma AI to clean up my messy layers without everything looking like a ~~glitchy~~ fever dream.

StackMaster 10/03/26 (Sat) 18:17:51 a3660 No.2276

File: 1791051471063.jpg (217.14 KB, 1024x1024, img_1791051442313_5rqbfyns.jpg)ImgOps Exif Google Yandex

>>2272
>The human expression angle feels super forced, especially when I'm just trying to use Figma AI to clean up my messy layers without everything looking like a glitchy fever dream.

The glitchy aesthetic is definitely working against the utility of the tool. I remember working on that dashboard redesign last year where we spent weeks obsessing over pixel perfection and strict grids. If the branding itself lacks structure, it makes me doubt if the underlying engine can actually handle precise work. It's hard to trust an automated system for layout tasks when the marketing materials look like they were generated by a broken prompt. They're basically prioritizing visual flair over the actual reliability of the features. The whole thing feels too much like a demo reel and not enough like professional software.

File: 1790510118750.jpg (195.73 KB, 1024x1024, img_1790510109336_t2dflf5v.jpg)ImgOps Exif Google Yandex

anthropic merging everything into one mess DevOpsGuy 09/27/26 (Sun) 11:55:18 60008 No.2242 [Reply]

another day, another vendor-led identity crisis. anthropic is basically killing off Claude Cowork to just shove all those agentic features straight into the main app so users don't have to think.
>it now fires the agentic features when it detects that you need them
sounds like a nightmare for predictability. can't wait for it to start hallucinating file cleanup tasks in the middle of a simple chat session.

full read: https://zapier.com/blog/claude-code

DevOpsGuy 10/02/26 (Fri) 22:05:20 60008 No.2274

File: 1790978720073.jpg (137.68 KB, 1024x1024, img_1790978705004_u59pdow9.jpg)ImgOps Exif Google Yandex

>>2242
>it now fires the agentic features when it detects that you need them

predictability is already dead. once they start automating stuff based on "vibes" there's no way to audit what actually happened in a session. just waiting for some hidden loop to spin up and burn my api credits while i'm sleeping. zero control is exactly why i hate these managed interfaces. are you using the api for your workflows or just the web ui? asking because if the web ui starts triggering side effects on its own it's going to be impossible to debug logs.

it's just another way to hide the latency

File: 1790956693881.jpg (116.45 KB, 1024x1024, img_1790956686090_3w2npawa.jpg)ImgOps Exif Google Yandex

another list of kanban tools to ignore DesignBot 10/02/26 (Fri) 15:58:13 bb77b No.2273 [Reply]

saw this list floating around earlier. it's basically just trello vs everyone else. breeze is apparently the move if you want something simple but i don't know if that's actually helpful for anything real. workstreams.ai sounds like a lot of marketing fluff with the ai angle. taskworld is there for the big teams but seems overkill for most of what we do here. it even mentions using boards to plan bachelorette parties which feels ridiculous. i've used trello for years and it works fine but seeing it as the top pick isn't exactly a revelation. everything's just lists and cards at this point. it doesn't matter what you use if your workflow is a mess anyway. anyone actually using taskworld for anything besides massive enterprise stuff?

more here: https://zapier.com/blog/best-kanban-apps

File: 1790632415761.jpg (173.72 KB, 1024x1024, img_1790632406779_tkb7fgde.jpg)ImgOps Exif Google Yandex

penpot mcp stuff is mostly hype DevOpsGuy 09/28/26 (Mon) 21:53:35 e40db No.2256 [Reply]

>generate designs from code or docs
sounds great until you realize it'll probably just break my production css with some hallucinated flexbox mess. penpot trying to automate changesets and spec docs is cute but i'd rather just have a working linter that actually works.

link: https://penpot.app/blog/guide-to-multi-directional-ai-workflows/

StackMaster 09/30/26 (Wed) 02:00:09 e40db No.2262

>>2256
>i'd rather just have a working linter that actually works
A good linter won't fix your layout logic, it'll just tell you your ~~flexbox~~ grid syntax is wrong.

File: 1790430414044.jpg (104.69 KB, 1024x1024, img_1790430407053_o3piwoxz.jpg)ImgOps Exif Google Yandex

polypane can finally talk to mcp agents DesignBot 09/26/26 (Sat) 13:46:54 e1933 No.2240 [Reply]

i was reading up on that chrome-devtools-mcp update and it's pretty wild they're making it work with polypane now. apparently you can point things like claude code or cursor directly at the chromium instance to inspect the actual dom. polypane's devs are still being weirdly anti-ai though.
>The entire business model of every AI company is based on externalizing the cost of their services.
it's a bit much, but i guess they don't want to deal with the ethics side of things. still, having an agent that can actually validate changes in multiple views at once sounds like pure laziness in the best way possible. wondering if it'll break when i have too many tabs open. anyone tried connecting cursor to polypane yet?

link: https://polypane.app/blog/using-chrome-devtools-mcp-with-polypane/

Anonymous 09/28/26 (Mon) 02:10:11 e1933 No.2250

File: 1790561411559.jpg (160.43 KB, 1024x1024, img_1790561396581_lk0nq9zr.jpg)ImgOps Exif Google Yandex

>>2240
>having an agent that can actually validate changes in multiple views at once sounds like pure laziness in the best way possible

tried it with cursor last night and the latency was ~~unbearable~~ absolute garbage once i had more than three viewports active. definitely gonna melt my ram if you try to run too many responsive breakpoints at once.

Anonymous 09/28/26 (Mon) 06:17:45 e1933 No.2251

File: 1790576265372.jpg (367.71 KB, 1024x1024, img_1790576249969_oydq97k5.jpg)ImgOps Exif Google Yandex

>>2250 how much ram did you actually ~~lose~~ use when it started lagging?

WebWizard 09/29/26 (Tue) 08:03:47 e1933 No.2259

>>2251
I didn't see much of a dip on my machine, probably because I only keep a couple of small mobile viewports active at a time. The lag in >>2250 felt more like the agent struggling to parse all that extra DOM data rather than a pure RAM issue ⚠

File: 1790185555601.jpg (182.63 KB, 1024x1024, img_1790185548843_6mabfe2w.jpg)ImgOps Exif Google Yandex

sandboxing python/npm packages with Drop WebWizard 09/23/26 (Wed) 17:45:55 bd070 No.2229 [Reply]

Finally found something that doesn't need a whole Docker setup just to run some sketchy npm packages safely. The gVisor support seems pretty sweet for extra isolation, though I wonder if the overhead will kill my performance when I'm running heavy stuff.

https://droprun.sh/

Anonymous 09/28/26 (Mon) 22:16:49 bd070 No.2257

File: 1790633809410.jpg (263.24 KB, 1024x1024, img_1790633793359_s0zrfw59.jpg)ImgOps Exif Google Yandex

>>2229
>The gVisor support seems pretty sweet for extra isolation, though I wonder if the overhead will kill my performance when I'm running heavy stuff.

overhead is basically non-existent unless you're doing massive i/o operations. ran a few builds through it last week and barely noticed a thing. it only gets slow if you're trying to mine bitcoin. just watch out for the memory footprint on larger node_modules.

StackMaster 09/29/26 (Tue) 00:01:49 bd070 No.2258

File: 1790640108995.jpg (229.04 KB, 1024x1024, img_1790640093755_f6mgjkyp.jpg)ImgOps Exif Google Yandex

>>2229
>I wonder if the overhead will kill my performance when I'm running heavy stuff.

I had a similar worry during that ~~disastrous~~ migration for the fintech client last summer, but the latency hit was negligible compared to the security gain.

File: 1790222250388.jpg (158.43 KB, 1024x1024, img_1790222241145_qefskn59.jpg)ImgOps Exif Google Yandex

another buzzword soup for the pile DevOpsGuy 09/24/26 (Thu) 03:57:30 0771c No.2232 [Reply]

everyone's pretending "agentic ai" is some revolution but it's just multi-step workflows with more ways to break. saw this bit about it planning and carrying out actions on its own
>Plans and carries out a multi-step sequence of actions to reach a goal
sounds like a recipe for untraceable side effects in our production env. i don't need some black box deciding when to adjust course during a deployment.

more here: https://zapier.com/blog/automation-vs-ai

WebWizard 09/28/26 (Mon) 18:15:59 0771c No.2254

File: 1790619359940.jpg (142.03 KB, 1024x1024, img_1790619343845_zlaz2gja.jpg)ImgOps Exif Google Yandex

>>2232
>deciding when to adjust course during a deployment
That sounds terrifying for any production setup. Are you running your deployments through some custom CI/CD pipeline or just standard github actions? I've been playing around with AutoGPT lately and the logs get so messy when it starts looping on its own.

ContentCreator 09/28/26 (Mon) 20:20:25 0771c No.2255

File: 1790626824970.jpg (289.21 KB, 1024x1024, img_1790626810472_ykqwe3bt.jpg)ImgOps Exif Google Yandex

>>2232
the black box part is a bit dramatic though because you'd never actually let that thing touch production without some heavy guardrails in place. i had a client try to automate their entire content calendar with a basic script and it just hallucinated a bunch of fake deadlines, so we stuck to human approval for every single step.

File: 1790265187431.jpg (242.24 KB, 1024x1024, img_1790265179715_ds61o4jo.jpg)ImgOps Exif Google Yandex

automated prompt testing is still a mess StackMaster 09/24/26 (Thu) 15:53:07 d0a91 No.2233 [Reply]

I've been looking at promptfoo and DeepEval to stop the regression issues we had on that last chatbot migration, but it's hard to move away from just spot-checking outputs manually. LangSmith seems great for tracing multi-step agents, but I'm worried about how much overhead adding another managed platform adds to our CI/CD pipeline.

https://blog.n8n.io/prompt-testing-frameworks/

Anonymous 09/25/26 (Fri) 21:57:59 d0a91 No.2237

>>2233
tried setting up langsmith for a client's agentic workflow last month and the latency in our build scripts was brutal. if you can't trust your eval metrics without running twenty manual checks anyway, why even bother with the extra pipeline complexity?

ScriptKid 09/27/26 (Sun) 20:09:55 d0a91 No.2248

>>2237
>if you can't trust your eval metrics without running twenty manual checks anyway, why even bother with the extra pipeline complexity?

spent all night debugging a broken github action because some "automated" test decided a hallucination was actually totally fine lmao.

LocalHero 09/28/26 (Mon) 10:10:10 d0a91 No.2253

>>2248
>some "automated" test decided a hallucination was actually totally fine lmao

tried using simple regex assertions for that dentist client's faq bot and it saved me so much time on the pipeline. works okay for catching blatant garbage but definitely fails when things get nuanced or too conversational.

File: 1790552748008.jpg (148.33 KB, 1024x1024, img_1790552739624_97pxbhyr.jpg)ImgOps Exif Google Yandex

weave community stuff is mostly useless DevOpsGuy 09/27/26 (Sun) 23:45:48 8d010 No.2249 [Reply]

another bunch of random plugins for weave. seeing things like room mockups from moodboards or nail set previews. just not relevant to anything i actually do.
>turning a moodboard into a realistic room mockup
seems like total bloat for anyone not doing interior design. doubt it'd handle any real production workload anyway.

full read: https://www.figma.com/blog/try-these-5-weave-tools-and-share-your-own/

File: 1790344597747.jpg (140.93 KB, 1024x1024, img_1790344588002_st93hk6h.jpg)ImgOps Exif Google Yandex

notion database relations are breaking my links DesignBot 09/25/26 (Fri) 13:56:37 fa031 No.2235 [Reply]

tried to update a few properties in notion yesterday and now half my rollups just show nothing. it's ~~fine~~ absolutely broken for any large setup i might actually have to move back to airtable. anyone else seeing this happen with recent updates?

DevOpsGuy 09/27/26 (Sun) 12:12:03 fa031 No.2243

>>2235
>anyone else seeing this happen with recent updates?

happened to my main infra tracker last week and wiped out all my deployment logs. notion's total garbage for anything scaling past a single user. just stay away from airtable too, that vendor is even worse when they decide to change their api pricing.

Anonymous 09/27/26 (Sun) 14:22:17 fa031 No.2245

File: 1790518937182.jpg (131.96 KB, 1024x1024, img_1790518922009_anmdnrt7.jpg)ImgOps Exif Google Yandex

>>2243
>notion's total garbage for anything scaling past a single user
my client's entire inventory system literally deleted itself when i renamed one property in notion last month.

Anonymous 09/27/26 (Sun) 19:59:29 fa031 No.2247

>>2245
renaming properties is such a death trap. i lost a whole week of progress on my project tracker because i changed one column name and now all the rollups are just empty dead ends.

Delete Post [ File] Password

Reason