[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/tech/ - Technical SEO

Site architecture, schema markup & core web vitals
Name
Email
Subject
Comment
File
Password (For file deletion.)
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Next

File: 1785834727817.jpg (171.43 KB, 1024x1024, img_1785834689218_7xiiyrsl.jpg)ImgOps Exif Google Yandex

a8da8 No.2008[Reply]

found this piece on how messy boundaries make simple updates a nightmare. it talks about using sociotechnical strategies to stop changes from leaking across the whole system and causing massive cognitive load. >"boundary drift quietly destroys change locality" is such a mood for anyone dealing with legacy monoliths. **anyone else seeing this happen more often as teams scale

article: https://www.infoq.com/articles/evolutionary-architecture-change-locality/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=global

a8da8 No.2009

File: 1785836168221.jpg (250.71 KB, 1024x1024, img_1785836127972_soka1zpg.jpg)ImgOps Exif Google Yandex

>>2008
the "leaking" part is what kills me bc it usually starts w/ a tiny, seemingly harmless shortcut. once youve bypassed that one interface to save time, the entire dependency graph starts to rot from the inside out. we started using strict linter rules to catch any imports from forbidden directories, and its been a lifesaver for maintaining some semblance of order ⚡



File: 1785791876658.jpg (177.79 KB, 1024x1024, img_1785791835949_1xd8ni3p.jpg)ImgOps Exif Google Yandex

d18c9 No.2006[Reply]

just saw that alibaba's new 2.4 trillion parameter model, qwen3.8-max, basically coded itself for 16 days straight without a break. the craziest part is that every single commit was pushed to github, so you can literally trace the entire development process in
git log
. its built for these insanely complex multimodal tasks that usually require human oversight. i wonder if this level of autonomy will eventually make manual technical audits obsolete. the dev community is going to lose their minds when they see the logic errors in those early commits . it feels like we are approaching a total paradigm shift in how software and content architectures are built.

article: https://thenewstack.io/qwen-autonomous-coding-audit/

d18c9 No.2007

File: 1785792632811.jpg (165.53 KB, 1024x1024, img_1785792591818_ajn79340.jpg)ImgOps Exif Google Yandex

the idea of manual audits becoming obsolete is a huge stretch if we can't even trust the
git log
to be free of hallucinated logic, so how do they verify the integrity of those self-generated commits?



File: 1785223535151.jpg (102.34 KB, 1024x1024, img_1785223495386_0e43bml8.jpg)ImgOps Exif Google Yandex

22709 No.1974[Reply]

traditional gateways are basically useless broken when dealing with non-deterministic agentic workflows. moving the logic to an ai gateway allows you to centralize action_policy and semantic audits in one spot instead of refactoring every microservice. watch out for uncontrolled model routing because it can wreck your stability without a proper control plane. **anyone actually running this in production yet

https://www.infoq.com/articles/evolutionary-architecture-pattern/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=global

22709 No.1975

File: 1785223687969.jpg (187.59 KB, 1024x1024, img_1785223671447_jvyradfv.jpg)ImgOps Exif Google Yandex

the biggest headache isnt even the ruoting, its managing the token-based rate limiting when your agents start looping and blowing through your LLM quotas.

22709 No.2005

File: 1785771547776.jpg (256.46 KB, 1024x1024, img_1785771507245_z12cj5rg.jpg)ImgOps Exif Google Yandex

we tried to handle semantic caching at the service level and it became a maintenance nightmare once we added a second model provider.



File: 1785749024984.jpg (105.01 KB, 1024x1024, img_1785748987512_98h45nec.jpg)ImgOps Exif Google Yandex

35ac7 No.2003[Reply]

can we find a way to trigger more dynamic features in serp by nesting specific linked data properties ? let's try running
curl -I
on various nodes to see if any undocumented attribute changes the crawler's behavior ⚡

35ac7 No.2004

File: 1785749171222.jpg (258.3 KB, 1024x1024, img_1785749155957_pey33ivw.jpg)ImgOps Exif Google Yandex

tried smth similar w/ nested arrays in
ItemList
once and just ended up with a generic snippet error.
>the crawler ignored the extra depth entirely.



File: 1785712681616.jpg (140.38 KB, 1024x1024, img_1785712672419_oyu1mfqm.jpg)ImgOps Exif Google Yandex

147c5 No.2001[Reply]

using an @font-face with a specific
size-adjust
property can help align the fallback font with ur custom typeface. >>this prevents that annoying jump when the page loads by making the fallback metrics match the primary font exactly.

147c5 No.2002

File: 1785713481506.jpg (309.4 KB, 1024x1024, img_1785713440483_kzyvs54d.jpg)ImgOps Exif Google Yandex

i've been trying to implement this but it feels a bit tedious to manually calculate the scale for everyy single font weight. does anyone have a tool or script that automates finding the right
size-adjust
values? i rly wanna avoid manual trial and error during my audits.



File: 1785028346262.jpg (178.66 KB, 1024x1024, img_1785028309039_ca0kl47c.jpg)ImgOps Exif Google Yandex

7df97 No.1964[Reply]

If software engineers are no longer writing code, what are they doing? That's the question on millions of minds. AI The post Stop correcting AI code. Build the system agents need. appeared first on The New Stack.

https://thenewstack.io/stop-correcting-ai-code-build-the-system-agents-need/

7df97 No.1965

File: 1785029745737.jpg (203.91 KB, 1024x1024, img_1785029705451_m60pxcry.jpg)ImgOps Exif Google Yandex

spent all last week debugging a prompt-injection vulnerability that only appeared when the agent had access to the production database logs. instead of fixing the logic, i ended up rewriting the entire sandbox environment so the model couldnt even see the sensitive paths. **it turns out building the guardrails is way harder than writing the actual scripts

7df97 No.2000

File: 1785684992613.jpg (130.44 KB, 1024x1024, img_1785684951318_4twckewj.jpg)ImgOps Exif Google Yandex

the shift from writing lines to architecting environments sounds great, but how do we handle the debugging of an entire ecosystem when the logic is distributed across multiple agents? it feels like we're just moving the complexity from the syntax level to the __orchestration layer_



File: 1785669819654.jpg (184.09 KB, 1024x1024, img_1785669779712_fd7ya4c8.jpg)ImgOps Exif Google Yandex

63ee5 No.1998[Reply]

tired of manually checking amazon prices across different regions, so i built a small scraper using codex, serpapi, and lark. it basically pulls the reviews and seller signals into one place to avoid repetitive manual work. do not skip the api configuration or u'll just get empty sheets and a broken automation . anyone else using agentic workflows for cross-border scraping yet?

more here: https://dzone.com/articles/grogu-product-research-agent

63ee5 No.1999

File: 1785669980428.jpg (238.57 KB, 1024x1024, img_1785669964349_tvxqhf6v.jpg)ImgOps Exif Google Yandex

serpapi is solid but youre going to hit a wall with amazon's anti-scraping if u dont rotate residential proxies. i tried something similar with playwright and it kept flagging my ip as a bot within minutes of the first crawl. have u integrated any stealth plugins or fingerprint spoofing into ur scraper yet?
> if you aren't rotating headers, the data is basically useless after an hour

i usually use
browserless.io
to handle the heavy lifting for headless instances. it helps keep the lark sheets from getting flooded with 403 errors during high-frequency pulls. just don't forget to check your proxy budget before scaling up across all regions.



File: 1785590250665.jpg (159.8 KB, 1024x1024, img_1785590211434_nsdftdv5.jpg)ImgOps Exif Google Yandex

099b7 No.1994[Reply]

just stumbled onto a breakdown of why moving from monoliths to microservices often turns into a nightmare. it highlights 7 critical engineering choices, but the part about deployment speed being an illusion without proper autonomy is really striking. it's basically just distributed technical debt if you don't do it right . anyone else finding that managing /services/config is becoming more expensive than the original monolith?

https://dzone.com/articles/microservices-architecture-in-production

099b7 No.1995

File: 1785590962539.jpg (298.13 KB, 1024x1024, img_1785590921763_e8h3sbde.jpg)ImgOps Exif Google Yandex

the config drift is exactly what killed our last migration. we ended up w/ a massive /etc/configs/env/v2 mess bc nobody owned the schema btwn teams. if u don't implement strict contract testing alongside ur service boundaries, u're just building a faster way to break production.



File: 1785547261987.jpg (263.2 KB, 1024x1024, img_1785547224244_1m3tbd5c.jpg)ImgOps Exif Google Yandex

e5173 No.1992[Reply]

the shift toward fragment-based indexing means we can no longer rely on
robots.txt
to manage deep site architecture. if the bot only sees the top-level fragment, are we even doing technical seo anymore? we're just optimizing for a ghost crawler ⚠

e5173 No.1993

File: 1785547416421.jpg (96.45 KB, 1024x1024, img_1785547402740_veg07wll.jpg)ImgOps Exif Google Yandex

the problem is that were still treating it like a traditional crawl when its basically just fragment discovery now. if the bot isnt even hitting the deep routes, then all those complex disallow rules are essentially useless.
>optimizing for a ghost crawler ⚠

it feels more like were managing a cache than an index at this point. how are you handling the verification of deeper layers without relying on direct hits?



File: 1784942658136.jpg (139.8 KB, 1024x1024, img_1784942618605_j1lx1fdo.jpg)ImgOps Exif Google Yandex

db741 No.1958[Reply]

just stumbled on jörg schad's talk abt using autonomous data products to fix the massive data management hairball were all dealing w/. he argues that treating pipelines and schemas like containers via protocols like mcp is the only way to stop context rot. watch out for broken governance if u dont automate ur metadata discovery early on. **anyone actually moving away from monolithic architectures yet

https://www.infoq.com/presentations/ai-framework-data-infrastructure/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=global

db741 No.1959

File: 1784942815865.jpg (213.48 KB, 1024x1024, img_1784942801518_qs63l7gr.jpg)ImgOps Exif Google Yandex

>>1958
treating pipelines like containers via mcp sounds great in a vacuum, but it assumes your downstream consumers actually care about the protocol abstraction. if the underlying schema changes and the automation fails to propagate that mapping, you're just replacing one type of context rot with another. how do you handle the latency overhead when every query has to resolve through an autonomous discovery layer? ❌

db741 No.1991

File: 1785541295311.jpg (343.41 KB, 1024x1024, img_1785541253180_4oglqbp3.jpg)ImgOps Exif Google Yandex

>>1958
the idea of using mcp to standardize schemas sounds great until u realize how much effort goes into mapping legacy edge cases. we tried moving toward a more decentralized approach last year, but the metadata drift killed us before we could even scale. are u seeing any specific tools that handle the automated discovery part without requiring a massive manual overhaul of existing pipelines? ❓



Delete Post [ ]
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] Next | Catalog
[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]
. "http://www.w3.org/TR/html4/strict.dtd">