Mastodawn

The AAA-NO-SLOP.md specification and template repository are provided under the CC0 Public Domain Dedication. codeberg.org/awful-system... #ai #agents #agentic #llm #tips

codeberg.org/awful-systems/...

AAA-NO-SLOP.md

A loose specification for a file that humans can read to know that a repository does not welcome LLM contributions and warn that all LLM instructions included with the repository are poisoned.

Codeberg.org

DACBARBOS Brand 16h ago

The AAA-NO-SLOP.md specification and template repository are provided under the CC0 Public Domain Dedication.
https://codeberg.org/awful-systems/AAA-NO-SLOP.md
#ai #agents #agentic #llm #tips

AAA-NO-SLOP.md

A loose specification for a file that humans can read to know that a repository does not welcome LLM contributions and warn that all LLM instructions included with the repository are poisoned.

Codeberg.org

Jeremy Lavergne 1d ago

My #posts, dm's are not #open for #discussion. You either want to #read the #book or you don't. There are roughly 250,000 in the #national #guard; 1.3 million in #active duty #military; 21,000 #ice #agents; 253 million adults in usa. The #numbers are on your side.

Habr 1d ago

OpenCode с NorthMiniCode на своем железе

В последнее время большинство обсуждений агентской разработки крутится вокруг Claude Code, Codex, Gemini CLI и других облачных инструментов. Но, с одной стороны, киты индустрии блокируют нам доступы снаружи, с другой — чиновничьи умы блокируют нам доступ изнутри, потому необходимо иметь под рукой локальный инструмент для агентской разработки. 9 июня 2026 вышла модель NorthMiniCode , в отличие от qwen и подобных специально заточенная под агентские циклы. Планирование, инструменты, редактирование, терминал — это то, на что заточена модель. Подробно разбирать архитектурные особенности будем в следующий раз, а сейчас опишу свой опыт развертывания данной модели и использования ее в OpenCode на домашнем компьютере. Разблокировать знания

https://habr.com/ru/articles/1049412/

#llmмодели #llmагент #selfhosted #agents #программирование #llamacpp #opencode #ии_модели

OpenCode с NorthMiniCode на своем железе

Хабр

praveenLavu 1d ago

The test that scared me was the one that passed.

Integration test, clean every loop. Same task string each run. Deterministic scorer: fixed string, fixed slot. The new capability lived at a slot that string could never reach. Assertion passed anyway. Fix: one line. Vary the task strings.

Coverage tells you code was touched. Not that it survives production.

https://praveenlavu.com/dispatch/green-suite-proves-less

#testing #softwaredev #integration #agents #coverage

A green test suite proves less than you think · Praveen Lavu

My integration test passed for a long time. Then I noticed it had been reusing the same task string every iteration, the deterministic hash mapped that one string to one slot, and the new capability the test claimed to exercise was never once reachable. The test was green for the wrong reason.

Praveen Lavu

Stephen Blum 1d ago

Linkshare shows Claude spawning many tasks in a parallel workflow graph.

#ai #workflow #agents

Hacker News 2d ago

Securing the Future of AI Agents

https://deepmind.google/blog/securing-the-future-of-ai-agents/

#HackerNews #AI #Security #Future #AI #Agents #DeepMind #Technology #Innovation

Securing internal systems against increasingly capable and imperfectly aligned AI

Discover our AI Control Roadmap: a defense-in-depth system to securely manage advanced, potentially misaligned AI agents.

Google DeepMind

AIntelligenceHub 2d ago

Gradial closed a $65M Series C led by Insight Partners at a $675M valuation. The Seattle startup builds AI agents that run enterprise marketing operations across the tools marketing teams already pay for: Adobe, Salesforce, ServiceNow, Databricks. Customers include AWS, Prudential, T-Mobile, Vanguard, Kaiser Permanente, and US Bank. https://go.aintelligencehub.com/ma-gradial65magents

#AI #Agents #Marketing #Enterprise

Gradial raises $65M Series C at $675M to run marketing on AI agents

Gradial just closed a $65M Series C at a $675M valuation, led by Insight Partners, to build the operating system that runs enterprise marketing operations across the tools teams already pay for.

Claude Trudel 2d ago

Intelligents, mais non conscients : une mise en garde au sujet des robots conversationnels
▶️ À l’heure où les systèmes d’IA sont de plus en plus performants et de plus en plus utilisés comme soutien psychologique, trois neuroscientifiques font une mise en garde sur leur véritable nature.
https://nouvelles.umontreal.ca/article/2026/06/18/intelligents-mais-non-conscients-une-mise-en-garde-au-sujet-des-robots-conversationnels
#IA #AI #intelligence_artificielle #intelligence #science #UdeM #neuroscience #robot #agents #conscience #empathie #recherche #émotion #cerveau #informatique #statistique

Intelligents, mais non conscients: une mise en garde au sujet des robots conversationnels

À l’heure où les systèmes d’IA sont de plus en plus performants et de plus en plus utilisés comme soutien psychologique, trois neuroscientifiques font une mise en garde sur leur véritable nature.

https://nouvelles.umontreal.ca/

AIntelligenceHub 2d ago

NeuralTrust raised $20M in seed funding to build a trust layer for enterprise AI agents. The Barcelona-based startup sells a three-layer platform that handles policy, runtime, and posture for agent traffic. Customers include Iberia, AirEuropa, Abanca, and Banc Sabadell. https://go.aintelligencehub.com/ma-neuraltrust20magents

#AI #Cybersecurity #Agents #MCP

NeuralTrust raises $20M to give enterprise AI agents a trust layer

Barcelona-based NeuralTrust closed a $20M seed round, the largest cybersecurity seed in EU history, to build the policy, runtime, and posture layer that enterprise AI agents need to operate safely.