#agenticcoding — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #agenticcoding, aggregated by home.social.
-
vibe coders crying "but I put so much effort into that! You could at least acknowledge that". You didn't put any effort into that. Your #LLM did. Pass my thanks to your LLM.
#AI #artificialIntelligence #vibeCoding #genAI #generativeAI #noAI #codingAgent #agenticAI #agenticCoding #promptEngineering
-
vibe coders crying "but I put so much effort into that! You could at least acknowledge that". You didn't put any effort into that. Your #LLM did. Pass my thanks to your LLM.
#AI #artificialIntelligence #vibeCoding #genAI #generativeAI #noAI #codingAgent #agenticAI #agenticCoding #promptEngineering
-
RT @MiaAI_lab: Reality-Check: Der neue Muse Glimmer 30b ist im agentic Coding nicht besser als Qwen3.6 27b. Und Qwen3.8 27b wird voraussichtlich morgen veröffentlicht.
mehr auf Arint.info
#AgenticCoding #AIUpdates #LLMComparison #MuseGlimmer #Qwen3 #arint_info
-
RT @MiaAI_lab: Reality-Check: Der neue Muse Glimmer 30b ist im agentic Coding nicht besser als Qwen3.6 27b. Und Qwen3.8 27b wird voraussichtlich morgen veröffentlicht.
mehr auf Arint.info
#AgenticCoding #AIRelease #LLMUpdate #MuseGlimmer #Qwen3 #arint_info
-
🚀 FCAIC #27 – Google Antigravity Skills in Action
Randal Schwartz walks through two custom skills:
🎨 Adventure-GM: stateful text and visual adventures, persistent logs, scene continuity & hidden puzzle state
⚙️ Ticket workflow: specialized agents, human approval gates, 5-pillar reviews & post-merge meta-learning📅 11 August 2026
⏰ 16:00 UTC / 09:00 PDT
🎥 https://tinyurl.com/FCAICYouTube -
🚀 FCAIC #27 – Google Antigravity Skills in Action
Randal Schwartz walks through two custom skills:
🎨 Adventure-GM: stateful text and visual adventures, persistent logs, scene continuity & hidden puzzle state
⚙️ Ticket workflow: specialized agents, human approval gates, 5-pillar reviews & post-merge meta-learning📅 11 August 2026
⏰ 16:00 UTC / 09:00 PDT
🎥 https://tinyurl.com/FCAICYouTube -
From the Leanpub Blog: Leanpub Book LAUNCH 🚀 Harness Engineering: Building Reliable Workflows Around Non-Deterministic Agents by Ian Johnson
#books #leanpublishing #selfpublishing #harnessengineering #aiagents #agenticcoding #llmreliability #devtools
-
From the Leanpub Blog: Leanpub Book LAUNCH 🚀 Harness Engineering: Building Reliable Workflows Around Non-Deterministic Agents by Ian Johnson
#books #leanpublishing #selfpublishing #harnessengineering #aiagents #agenticcoding #llmreliability #devtools
-
Adding a layer of indirection solves most problems 🤓:
“Managing AI Coding Costs At Scale”, Databricks (https://www.databricks.com/blog/managing-ai-coding-costs-scale).
Via HN: https://news.ycombinator.com/item?id=49214468
#AI #SoftwareEngineering #Coding #AIAssistedCoding #Costs #AgenticCoding #Harness
-
Adding a layer of indirection solves most problems 🤓:
“Managing AI Coding Costs At Scale”, Databricks (https://www.databricks.com/blog/managing-ai-coding-costs-scale).
Via HN: https://news.ycombinator.com/item?id=49214468
#AI #SoftwareEngineering #Coding #AIAssistedCoding #Costs #AgenticCoding #Harness
-
I think Micah Lee is a big too much acerbic. Yes, the AI bubble is full of hype and smoke mirrors. Yes, the AI bubble will deflate. In any case, I do indeed believe that LLMs will get cheaper - way too cheaper form American companies. Why? Because the Chinese State will continue to fund local AI companies to produce even more efficient and faster models. Eventually, the whole industry will most likely be commoditized... Oh, and I guess the remaining Chinese IT companies will become bigger and bigger.
"The AI industry is a bubble built on hype and lies and debt, and the promise that executives can fire all their workers and still make lots of profit. It's a house of mirrors that will inevitably come crashing down. Anthropic and OpenAI sell $20/month plans knowing that if their users use as many tokens as they're given, it costs them way more money than it makes them. But they don't care because everything is fake. It's all terrible, and at some point soon, the whole thing will collapse.
I have no idea exactly what it will look like when that happens, but I do know that the era of cheap access to extremely powerful LLMs will end. It's only a matter of time before the actual costs will catch up to them. This already happened with GitHub Copilot. Starting in June, GitHub began charging based on actual usage, and it became way too expensive, so I (and many, many others) stopped using it.
Given this, I've decided that since agentic coding is able to write high quality code (if done correctly, not just by default lol) much faster than I can, and since frontier models are cheaper now than they ever will be again, I might as well take advantage of them while I can.
When everything crashes back down to earth, destroying the economy with it, hopefully we'll be left with some capable open models. And with our brains intact, too, of course."
https://micahflee.com/agentic-coding-techniques/
#AI #GenerativeAI #LLMs #AIAgents #AgenticCoding #SoftwareDevelopment #Programming
-
I think Micah Lee is a big too much acerbic. Yes, the AI bubble is full of hype and smoke mirrors. Yes, the AI bubble will deflate. In any case, I do indeed believe that LLMs will get cheaper - way too cheaper form American companies. Why? Because the Chinese State will continue to fund local AI companies to produce even more efficient and faster models. Eventually, the whole industry will most likely be commoditized... Oh, and I guess the remaining Chinese IT companies will become bigger and bigger.
"The AI industry is a bubble built on hype and lies and debt, and the promise that executives can fire all their workers and still make lots of profit. It's a house of mirrors that will inevitably come crashing down. Anthropic and OpenAI sell $20/month plans knowing that if their users use as many tokens as they're given, it costs them way more money than it makes them. But they don't care because everything is fake. It's all terrible, and at some point soon, the whole thing will collapse.
I have no idea exactly what it will look like when that happens, but I do know that the era of cheap access to extremely powerful LLMs will end. It's only a matter of time before the actual costs will catch up to them. This already happened with GitHub Copilot. Starting in June, GitHub began charging based on actual usage, and it became way too expensive, so I (and many, many others) stopped using it.
Given this, I've decided that since agentic coding is able to write high quality code (if done correctly, not just by default lol) much faster than I can, and since frontier models are cheaper now than they ever will be again, I might as well take advantage of them while I can.
When everything crashes back down to earth, destroying the economy with it, hopefully we'll be left with some capable open models. And with our brains intact, too, of course."
https://micahflee.com/agentic-coding-techniques/
#AI #GenerativeAI #LLMs #AIAgents #AgenticCoding #SoftwareDevelopment #Programming
-
RE: https://infosec.exchange/@isotopp/116987998761522351
Heute schenkt euch @isotopp einen lesenswerten Thread über agentisches Handwerkszeug und einen herrlich unkomplizierten Erklärbär-Artikel dazu. #LLM #AgenticCoding
-
RE: https://infosec.exchange/@isotopp/116987998761522351
Heute schenkt euch @isotopp einen lesenswerten Thread über agentisches Handwerkszeug und einen herrlich unkomplizierten Erklärbär-Artikel dazu. #LLM #AgenticCoding
-
Agentic Coding Meetup Bonn: KI für Soloselbstständige und kleine Teams - Praxis-Workflows für Buchhaltung, Backoffice Vertrieb & Co
Alte Kaffeerösterei, Donnerstag, 10. September um 17:00 MESZ
Offener Abend für alle, die mit Agentic Development (Claude Code, OpenCode, Codex, etc.) arbeiten oder einsteigen wollen. Alle Level: Anfänger bis Profis, kostenlos. Das Meetup ist ein gemeinsames Projekt mit pulsecode IT solutions .
Wir treffen uns donnerstags von 17 bis 19 Uhr in unseren Räumen in der Alten Kaffeerösterei in der Bonner Nordstadt.
Heute mit Vortrag: Philip Gabriel spricht über KI für Soloselbstständige und kleine Teams - Praxis-Workflows für Buchhaltung, Backoffice Vertrieb & Co
-
I’m considering open-sourcing Cue and Elicit if there’s interest. Cue: just-in-time skill/guide routing. Elicit: out-of-band requirements decisions. Would you use either? Reply with which one and your coding agent. That’ll help me decide what to package first. #iOSDev #AgenticCoding #SwiftDev
-
I’m considering open-sourcing Cue and Elicit if there’s interest. Cue: just-in-time skill/guide routing. Elicit: out-of-band requirements decisions. Would you use either? Reply with which one and your coding agent. That’ll help me decide what to package first. #iOSDev #AgenticCoding #SwiftDev
-
Tinkering with Pi in Docker – it’s surprisingly fun to build with your own agent!
This past month I've been diving head first into using a agentic coding harness and it's been a lot of fun! I decided to go trying out #pi from pi.dev for several reasons:
- I like the philosophy of building a coding agent the way *you* want it. Plus I love customization! 😁
- As a fellow tinkerer and builder I really like to understand how code and processes work behind the scenes; pi allows me to see the output as it builds.
- It's minimalistic. When you install it, it doesn't come with much of anything but a few basic tools. You can install (or even ask it to build!) the extensions, skills, prompts that you want. It also has significantly lower number system prompt tokens. compared to other popular coding agents too.
- You can use whatever models you want and as user of Local AI models I can easily hook up my models from llama.cpp.I'm really liking my experience with pi so far and I see why it's considered the *developer's* coding agent. If you are comfortable with code and the terminal, Pi is great tool to work with!
As I'm getting familar with coding using a coding agent harness like Pi, I got this running in a plain Docker container until I get use to it as I don't want to go accidently deleting important data or even my computer until I get the hang of it 😅
#ai #localai #pi #codingagent #buildinpublic #agentic #docker #llama #claude #codex #claudecode #coding #openclaw #agenticcoding
-
Tinkering with Pi in Docker – it’s surprisingly fun to build with your own agent!
This past month I've been diving head first into using a agentic coding harness and it's been a lot of fun! I decided to go trying out #pi from pi.dev for several reasons:
- I like the philosophy of building a coding agent the way *you* want it. Plus I love customization! 😁
- As a fellow tinkerer and builder I really like to understand how code and processes work behind the scenes; pi allows me to see the output as it builds.
- It's minimalistic. When you install it, it doesn't come with much of anything but a few basic tools. You can install (or even ask it to build!) the extensions, skills, prompts that you want. It also has significantly lower number system prompt tokens. compared to other popular coding agents too.
- You can use whatever models you want and as user of Local AI models I can easily hook up my models from llama.cpp.I'm really liking my experience with pi so far and I see why it's considered the *developer's* coding agent. If you are comfortable with code and the terminal, Pi is great tool to work with!
As I'm getting familar with coding using a coding agent harness like Pi, I got this running in a plain Docker container until I get use to it as I don't want to go accidently deleting important data or even my computer until I get the hang of it 😅
#ai #localai #pi #codingagent #buildinpublic #agentic #docker #llama #claude #codex #claudecode #coding #openclaw #agenticcoding
-
Agentic Coding Meetup Bonn
Alte Kaffeerösterei, Donnerstag, 30. Juli um 17:00 MESZ
Offener Abend für alle, die mit Agentic Development (Claude Code, OpenCode, Codex, etc.) arbeiten oder einsteigen wollen. Alle Level: Anfänger bis Profis, kostenlos. Das Meetup ist ein gemeinsames Projekt mit pulsecode IT solutions .
Wir treffen uns donnerstags von 17 bis 19 Uhr in unseren Räumen in der Alten Kaffeerösterei in der Bonner Nordstadt.
Heute mit Vortrag: Askan Schmeisser zeigt seinen kompletten Self-Hosting-Stack für AI-Agenten: Mattermost als Chat-Interface für Hermes Agent und OpenClaw, n8n als Automatisierungs-Orchestrator, und Outline als Wiki, in dem Agenten-Output automatisch publiziert wird. Dazu zeigt er seine konkreten Workflows.
-
Why Google’s Gemini 3.5 Pro Is Facing Delays: Inside the AI Giant’s Biggest Challenge Yet—and Why OpenAI and Anthropic Are Pulling Ahead in the Agentic Coding Race
#AI #AIModel #AINews #Tech #TechNews #LLMs #AIRace #Gemini #GoogleAI #AgenticCoding #AIDevelopment
maniainc.com/technology/w...
Why Google’s Gemini 3.5 ... -
Why Google’s Gemini 3.5 Pro Is Facing Delays: Inside the AI Giant’s Biggest Challenge Yet—and Why OpenAI and Anthropic Are Pulling Ahead in the Agentic Coding Race
#AI #AIModel #AINews #Tech #TechNews #LLMs #AIRace #Gemini #GoogleAI #AgenticCoding #AIDevelopment
maniainc.com/technology/w...
Why Google’s Gemini 3.5 ... -
#DarkFactory in der SoftwareEntwicklung - nicht mehr ganz so weit weg. Kostet natürlich auch einiges. Aber schon interessant zu sehen, wo die Reise hingeht
https://slides.com/johann-peterhartmann/dark-factories#/0/13
-
#DarkFactory in der SoftwareEntwicklung - nicht mehr ganz so weit weg. Kostet natürlich auch einiges. Aber schon interessant zu sehen, wo die Reise hingeht
https://slides.com/johann-peterhartmann/dark-factories#/0/13
-
Warum bessere Prompts deinen KI-Code nicht retten
https://blog.mayflower.de/30570-reward-hacking.html
> Die einen feiern das Tempo: Features in Stunden statt Tagen, ganze Services aus einer Handvoll Sätzen. Die anderen zeigen auf die Trümmer dahinter: aufgeblähter, schwer wartbarer Code, den am Ende doch wieder ein Mensch geradeziehen muss.