Stop "vibe coding" with AI agents in Flutter! 🤖 Learn the Synthetic Scar Architecture from @randalschwartz.bsky.social to give autonomous agents battle-tested, production-grade reflexes.👇
📅 10 Nov @ 17:00 UTC / 9:00 PT
🎥 tinyurl.com/FCAICYouTube
#FCAIC #FlutterDev #AgenticCoding
📅 10 Nov @ 17:00 UTC / 9:00 PT
🎥 tinyurl.com/FCAICYouTube
#FCAIC #FlutterDev #AgenticCoding
November 8, 2026 at 11:39 PM
KI-Agenten haben in 30 Stunden 948 Wege gefunden, auf Linux Admin-Rechte zu erlangen – laut Berichten für rund 1.300 $ pro Lücke, inklusive Korrekturvorschlägen.
Agenten schreiben Code – und lesen ihn schneller als jedes Prüf-Team.
#AgenticCoding #ITSicherheit
Agenten schreiben Code – und lesen ihn schneller als jedes Prüf-Team.
#AgenticCoding #ITSicherheit
November 9, 2026 at 8:43 AM
KI-Agenten haben in 30 Stunden 948 Wege gefunden, auf Linux Admin-Rechte zu erlangen – laut Berichten für rund 1.300 $ pro Lücke, inklusive Korrekturvorschlägen.
Agenten schreiben Code – und lesen ihn schneller als jedes Prüf-Team.
#AgenticCoding #ITSicherheit
Agenten schreiben Code – und lesen ihn schneller als jedes Prüf-Team.
#AgenticCoding #ITSicherheit
Nicht das Modell macht den Unterschied, sondern das Gerüst drumherum: der Agent Harness – Werkzeuge, Gedächtnis, Arbeitsschleife, Rechte.
Laut BairesDev schreibt KI bei 42 % der Entwickler mindestens die Hälfte des Codes (Vorjahr 12 %).
#KI #AgenticCoding
Laut BairesDev schreibt KI bei 42 % der Entwickler mindestens die Hälfte des Codes (Vorjahr 12 %).
#KI #AgenticCoding
November 8, 2026 at 8:40 PM
Nicht das Modell macht den Unterschied, sondern das Gerüst drumherum: der Agent Harness – Werkzeuge, Gedächtnis, Arbeitsschleife, Rechte.
Laut BairesDev schreibt KI bei 42 % der Entwickler mindestens die Hälfte des Codes (Vorjahr 12 %).
#KI #AgenticCoding
Laut BairesDev schreibt KI bei 42 % der Entwickler mindestens die Hälfte des Codes (Vorjahr 12 %).
#KI #AgenticCoding
Mythos: Coding-Agenten sind nur Spielerei.
Laut GitHubs Octoverse 2026 starteten KI-Agenten im September über 40 % aller Pull Requests (rund 110 Mio.), im März waren es 17 %.
Gestartet heißt nicht gut – das Review entscheidet.
Quelle: GitHub Octoverse, via Visual Studio Magazine #AgenticCoding
Laut GitHubs Octoverse 2026 starteten KI-Agenten im September über 40 % aller Pull Requests (rund 110 Mio.), im März waren es 17 %.
Gestartet heißt nicht gut – das Review entscheidet.
Quelle: GitHub Octoverse, via Visual Studio Magazine #AgenticCoding
November 9, 2026 at 9:42 PM
Mythos: Coding-Agenten sind nur Spielerei.
Laut GitHubs Octoverse 2026 starteten KI-Agenten im September über 40 % aller Pull Requests (rund 110 Mio.), im März waren es 17 %.
Gestartet heißt nicht gut – das Review entscheidet.
Quelle: GitHub Octoverse, via Visual Studio Magazine #AgenticCoding
Laut GitHubs Octoverse 2026 starteten KI-Agenten im September über 40 % aller Pull Requests (rund 110 Mio.), im März waren es 17 %.
Gestartet heißt nicht gut – das Review entscheidet.
Quelle: GitHub Octoverse, via Visual Studio Magazine #AgenticCoding
#ThinkingMachines‘ #Inkling2Small, a 594B-parameter Mixture-of-Experts model with 17B active, supports image inputs and variable effort via Tinker, with Apache 2.0 weights planned after safety testing. It improves #agenticcoding, #vision, #knowledgework and #reasoning, enabling autonomous tasks…
tech news ᳇ eicker.news (@technews@eicker.news)
#ThinkingMachines‘ #Inkling2Small, a 594B-parameter Mixture-of-Experts model with 17B active, supports image inputs and variable effort via Tinker, with Apache 2.0 weights planned after safety testing. It improves #agenticcoding, #vision, #knowledgework and #reasoning, enabling autonomous tasks like city simulation and time-series forecasting. https://thinkingmachines.ai/news/inkling-2-small/?eicker.news #tech #news #ainews
eicker.news
November 7, 2026 at 2:42 PM
#ThinkingMachines‘ #Inkling2Small, a 594B-parameter Mixture-of-Experts model with 17B active, supports image inputs and variable effort via Tinker, with Apache 2.0 weights planned after safety testing. It improves #agenticcoding, #vision, #knowledgework and #reasoning, enabling autonomous tasks…
Stop "vibe coding" with AI agents in Flutter! 🤖 Learn the Synthetic Scar Architecture from @randalschwartz.bsky.social to give autonomous agents battle-tested, production-grade reflexes.👇
📅 10 Nov @ 17:00 UTC / 9:00 PT
🎥 tinyurl.com/FCAICYouTube
#FCAIC #FlutterDev #AgenticCoding
📅 10 Nov @ 17:00 UTC / 9:00 PT
🎥 tinyurl.com/FCAICYouTube
#FCAIC #FlutterDev #AgenticCoding
November 8, 2026 at 11:38 PM
Stop "vibe coding" with AI agents in Flutter! 🤖 Learn the Synthetic Scar Architecture from @randalschwartz.bsky.social to give autonomous agents battle-tested, production-grade reflexes.👇
📅 10 Nov @ 17:00 UTC / 9:00 PT
🎥 tinyurl.com/FCAICYouTube
#FCAIC #FlutterDev #AgenticCoding
📅 10 Nov @ 17:00 UTC / 9:00 PT
🎥 tinyurl.com/FCAICYouTube
#FCAIC #FlutterDev #AgenticCoding
Check out the Ripple‑65B preview – a fully open Apache 2.0 US model that’s killing it at agentic coding. The benchmarks look insane, and the code is all yours. Dive in for the full scoop! #Ripple65B #Apache2 #AgenticCoding
🔗 aidailypost.com/news/ripple-...
🔗 aidailypost.com/news/ripple-...
November 6, 2026 at 8:26 PM
Check out the Ripple‑65B preview – a fully open Apache 2.0 US model that’s killing it at agentic coding. The benchmarks look insane, and the code is all yours. Dive in for the full scoop! #Ripple65B #Apache2 #AgenticCoding
🔗 aidailypost.com/news/ripple-...
🔗 aidailypost.com/news/ripple-...
🙌 Community pick
The-Cartographer — Ask which agent skills fit a task before you start it. One model call reads your task against the whole SKILL.md catalogue (193 skills bundled, plus yours) and names the few worth…
#indiedev #opensource #python #agentskills #agenticcoding
📮 Submit yours →
The-Cartographer — Ask which agent skills fit a task before you start it. One model call reads your task against the whole SKILL.md catalogue (193 skills bundled, plus yours) and names the few worth…
#indiedev #opensource #python #agentskills #agenticcoding
📮 Submit yours →
The-Cartographer · SilentStars
Ask which agent skills fit a task before you start it. One model call reads your task against the whole SKILL.md catalogue (193 skills bundled, plus yours) and names the few worth reading. CLI and MCP server for Claude Code, Codex, the Anthropic API or a local model. Python, standard library only.
silentstars.dev
November 2, 2026 at 1:46 PM
🙌 Community pick
The-Cartographer — Ask which agent skills fit a task before you start it. One model call reads your task against the whole SKILL.md catalogue (193 skills bundled, plus yours) and names the few worth…
#indiedev #opensource #python #agentskills #agenticcoding
📮 Submit yours →
The-Cartographer — Ask which agent skills fit a task before you start it. One model call reads your task against the whole SKILL.md catalogue (193 skills bundled, plus yours) and names the few worth…
#indiedev #opensource #python #agentskills #agenticcoding
📮 Submit yours →
You should test MORE as you introduce new production process. You WANT to verify that the quality does not degrade during the transition. Or later, but then you know already where to focus on.
#AI #AgenticCoding #Quality #Cybersecurity
#AI #AgenticCoding #Quality #Cybersecurity
October 29, 2026 at 4:20 PM
You should test MORE as you introduce new production process. You WANT to verify that the quality does not degrade during the transition. Or later, but then you know already where to focus on.
#AI #AgenticCoding #Quality #Cybersecurity
#AI #AgenticCoding #Quality #Cybersecurity
최근에는 ChatGPT의 Work/Codex를 이용한 작업을 하다 보니 보안상 파일 시스템 격리(isolation)가 필요하다고 생각되어 확인해 보았습니다. 요즘은 Docker에서 Agentic Coding 등을 위한 Docker Sandboxes 기능을 지원하더라고요. 간단한 설정 등을 확인하고 테스트해 보았습니다. workbench.edamyrepo.com/2026/10/12/d...
#Docker #AgenticCoding
#Docker #AgenticCoding
Docker를 통한 Agentic Coding의 Sandbox화 – Ed's 지식 공방
workbench.edamyrepo.com
November 1, 2026 at 12:39 PM
최근에는 ChatGPT의 Work/Codex를 이용한 작업을 하다 보니 보안상 파일 시스템 격리(isolation)가 필요하다고 생각되어 확인해 보았습니다. 요즘은 Docker에서 Agentic Coding 등을 위한 Docker Sandboxes 기능을 지원하더라고요. 간단한 설정 등을 확인하고 테스트해 보았습니다. workbench.edamyrepo.com/2026/10/12/d...
#Docker #AgenticCoding
#Docker #AgenticCoding
So. Endlich aus der reading list abhaken können:
https://blog.mayflower.de/32673-ki-agenten-verifikation-pruefschleife.html
> KI-Agenten Verifikation: Spitzenmodelle lösen 12-Stunden-Aufgaben nur jedes zweite Mal. Warum die Prüfung außerhalb des Modells liegen muss.
#ai #genai #agenticcoding
https://blog.mayflower.de/32673-ki-agenten-verifikation-pruefschleife.html
> KI-Agenten Verifikation: Spitzenmodelle lösen 12-Stunden-Aufgaben nur jedes zweite Mal. Warum die Prüfung außerhalb des Modells liegen muss.
#ai #genai #agenticcoding
October 28, 2026 at 8:16 PM
So. Endlich aus der reading list abhaken können:
https://blog.mayflower.de/32673-ki-agenten-verifikation-pruefschleife.html
> KI-Agenten Verifikation: Spitzenmodelle lösen 12-Stunden-Aufgaben nur jedes zweite Mal. Warum die Prüfung außerhalb des Modells liegen muss.
#ai #genai #agenticcoding
https://blog.mayflower.de/32673-ki-agenten-verifikation-pruefschleife.html
> KI-Agenten Verifikation: Spitzenmodelle lösen 12-Stunden-Aufgaben nur jedes zweite Mal. Warum die Prüfung außerhalb des Modells liegen muss.
#ai #genai #agenticcoding
Mirendil launched Fern, its first model, pretrained from scratch with a context window of up to 10M tokens 🌿 That's about half a million lines of code in view at once, built for agentic coding, long logs and full agent traces.
#AI #LLM #AgenticCoding
#AI #LLM #AgenticCoding
Introducing Fern
Fern is Mirendil’s first model. It is pretrained from scratch and supports a context window of up to 10M tokens.
mirendil.com
October 26, 2026 at 5:24 PM
Mirendil launched Fern, its first model, pretrained from scratch with a context window of up to 10M tokens 🌿 That's about half a million lines of code in view at once, built for agentic coding, long logs and full agent traces.
#AI #LLM #AgenticCoding
#AI #LLM #AgenticCoding
🚀 Kimi K3.1 Pro just gave Claude Opus a run for its money in coding and reasoning. With a 1‑million‑token context and agentic coding tricks, Moonshot AI’s open‑weight model is a game‑changer. Dive into the benchmark showdown! #KimiK31Pro #ClaudeOpus #AgenticCoding
🔗 aidailypost.com/news/kimi-k3...
🔗 aidailypost.com/news/kimi-k3...
October 25, 2026 at 4:34 AM
🚀 Kimi K3.1 Pro just gave Claude Opus a run for its money in coding and reasoning. With a 1‑million‑token context and agentic coding tricks, Moonshot AI’s open‑weight model is a game‑changer. Dive into the benchmark showdown! #KimiK31Pro #ClaudeOpus #AgenticCoding
🔗 aidailypost.com/news/kimi-k3...
🔗 aidailypost.com/news/kimi-k3...
Tej's 5 levels of agentic coding is a management warning first: at level 4 his agents cleared 273 of 475 tickets in 3 days, but left 66,725 words of decisions and a 6-hour outage he spotted himself. Speed without understanding is cognitive debt. #AI #AgenticCoding #Leadership
The 5 Levels of Agentic Coding: Why I Mostly Stay at Level 2
I've built with 1 agent, 8 agents in parallel worktrees and a software factory that ran for days on its own. My sweet spot is level 2 or 3. Here's why.
tej.as
October 24, 2026 at 11:37 AM
Tej's 5 levels of agentic coding is a management warning first: at level 4 his agents cleared 273 of 475 tickets in 3 days, but left 66,725 words of decisions and a 6-hour outage he spotted himself. Speed without understanding is cognitive debt. #AI #AgenticCoding #Leadership
Vibe Coding vs Agentic Coding 🚀 Writing code is just the beginning! Discover how AI-driven development differs in execution, testing, security, and scalability—and why production-ready software requires more than code that works. #AI #VibeCoding #AgenticCoding
blogs.emorphis.com/vibe-coding-...
blogs.emorphis.com/vibe-coding-...
Vibe Coding vs Agentic Coding - Key Differences - blogs.emorphis
Compare vibe coding vs agentic coding, learn how AI coding affects software quality, integrations, and scalability.
blogs.emorphis.com
October 23, 2026 at 9:47 AM
Vibe Coding vs Agentic Coding 🚀 Writing code is just the beginning! Discover how AI-driven development differs in execution, testing, security, and scalability—and why production-ready software requires more than code that works. #AI #VibeCoding #AgenticCoding
blogs.emorphis.com/vibe-coding-...
blogs.emorphis.com/vibe-coding-...
3. The US has benefited from move-fast-and-break-things mentality and also suffered for it. See above #1.
4. Europe leads interest in #DDDesign but often fail to apply it well.
Given #GenAI or #AgenticCoding or the flavor of the week...
3/4
4. Europe leads interest in #DDDesign but often fail to apply it well.
Given #GenAI or #AgenticCoding or the flavor of the week...
3/4
October 20, 2026 at 5:33 AM
3. The US has benefited from move-fast-and-break-things mentality and also suffered for it. See above #1.
4. Europe leads interest in #DDDesign but often fail to apply it well.
Given #GenAI or #AgenticCoding or the flavor of the week...
3/4
4. Europe leads interest in #DDDesign but often fail to apply it well.
Given #GenAI or #AgenticCoding or the flavor of the week...
3/4
In short, do not beg the LLM to do stuff, do not think you're smarter than it, know that there are ways to check it, and remember it does NOT know the specifics of your projects: it very likely sees them for the first time.
#AI #AgenticCoding #CodingAgents #DevTools #WebDev
#AI #AgenticCoding #CodingAgents #DevTools #WebDev
October 19, 2026 at 12:04 PM
In short, do not beg the LLM to do stuff, do not think you're smarter than it, know that there are ways to check it, and remember it does NOT know the specifics of your projects: it very likely sees them for the first time.
#AI #AgenticCoding #CodingAgents #DevTools #WebDev
#AI #AgenticCoding #CodingAgents #DevTools #WebDev
A safety classifier is a filter, not a security boundary. If your coding agent runs in auto mode, check the sandbox settings before you trust the classifier.
www.convective.com/blog/a-clas... #AISecurity #AgenticCoding
www.convective.com/blog/a-clas... #AISecurity #AgenticCoding
Is Claude Code Auto Mode Safe? A Classifier Is Not a Sandbox | Convective
Claude Code auto mode's safety classifier is a filter, not a security boundary. Why a prompt injection got through, and the sandbox settings to check first.
www.convective.com
October 19, 2026 at 9:00 PM
A safety classifier is a filter, not a security boundary. If your coding agent runs in auto mode, check the sandbox settings before you trust the classifier.
www.convective.com/blog/a-clas... #AISecurity #AgenticCoding
www.convective.com/blog/a-clas... #AISecurity #AgenticCoding
Mittelständisches Unternehmen, Software-Team von vier Devs, alles Pros.
Wir machen sehr erfolgreich #agenticcoding #eu-Souverän OHNE #claude, #openai etc.
So geht's:
- #mistral #vibe mit GLM 5.3 Model (Mistral Large 4 reift gerade)
- Vibe via ACP in Editor einbinden (ZED) oder VS-Code […]
Wir machen sehr erfolgreich #agenticcoding #eu-Souverän OHNE #claude, #openai etc.
So geht's:
- #mistral #vibe mit GLM 5.3 Model (Mistral Large 4 reift gerade)
- Vibe via ACP in Editor einbinden (ZED) oder VS-Code […]
Original post on mastodon.social
mastodon.social
October 18, 2026 at 6:41 PM
Job-Angst wg. #agenticcoding?
Meine These: Wer ein(e) echte(r) Dev ist, wirklich Ahnung hat und die neuen Tools umarmt und nutzt, wird überleben. Und profitieren.
Ich bin seit 30+ Jahren im Geschäft. Habe viele Entwicklungen mitgemacht, und mich immer mit entwickelt. So auch jetzt.
Und seit […]
Meine These: Wer ein(e) echte(r) Dev ist, wirklich Ahnung hat und die neuen Tools umarmt und nutzt, wird überleben. Und profitieren.
Ich bin seit 30+ Jahren im Geschäft. Habe viele Entwicklungen mitgemacht, und mich immer mit entwickelt. So auch jetzt.
Und seit […]
Original post on mastodon.social
mastodon.social
October 18, 2026 at 6:49 PM
Job-Angst wg. #agenticcoding?
Meine These: Wer ein(e) echte(r) Dev ist, wirklich Ahnung hat und die neuen Tools umarmt und nutzt, wird überleben. Und profitieren.
Ich bin seit 30+ Jahren im Geschäft. Habe viele Entwicklungen mitgemacht, und mich immer mit entwickelt. So auch jetzt.
Und seit […]
Meine These: Wer ein(e) echte(r) Dev ist, wirklich Ahnung hat und die neuen Tools umarmt und nutzt, wird überleben. Und profitieren.
Ich bin seit 30+ Jahren im Geschäft. Habe viele Entwicklungen mitgemacht, und mich immer mit entwickelt. So auch jetzt.
Und seit […]
AWS describes deployment paths for Reflection AI’s Beam. Our view at HighTalk: open weights don’t settle codebase suitability or operating responsibility. Choose model and infrastructure together: who owns the environment and checks the output? #AgenticCoding #AIInfrastructure
October 18, 2026 at 2:25 AM
AWS describes deployment paths for Reflection AI’s Beam. Our view at HighTalk: open weights don’t settle codebase suitability or operating responsibility. Choose model and infrastructure together: who owns the environment and checks the output? #AgenticCoding #AIInfrastructure
Just what are we doing here, people? Find the full video on the @gracefuldev YouTube channel. zurl.co/9An3z #ai #agenticcoding
October 15, 2026 at 6:39 PM
Just what are we doing here, people? Find the full video on the @gracefuldev YouTube channel. zurl.co/9An3z #ai #agenticcoding
Meta's new Muse Spark Open runs on NVIDIA's GB300 NVL72 with Blackwell Ultra GPUs, offering a 1M-token window and 816B parameters. It's a mixture-of-experts engine that actually understands codebases and powers agentic coding workflows. #MuseSparkOpen #AgenticCoding #GB300NVL72
🔗
🔗
October 15, 2026 at 6:57 PM
Meta's new Muse Spark Open runs on NVIDIA's GB300 NVL72 with Blackwell Ultra GPUs, offering a 1M-token window and 816B parameters. It's a mixture-of-experts engine that actually understands codebases and powers agentic coding workflows. #MuseSparkOpen #AgenticCoding #GB300NVL72
🔗
🔗