Herd – a lightweight multi-agent IDE, built with GStack
Herd는 다중 AI 코딩 에이전트를 한 데서 관리할 수 있는 경량화된 멀티 에이전트 IDE이다. 사용자는 여러 에이전트를 병렬로 실행하고, 실시간으로 작업 진행 상황을 모니터링하며, 코드 리팩토링과 테스트를 동시에 수행할 수 있다. Claude Code, Codex 등 다양한 AI 에이전트와 호환되며, GStack과 같은 스킬을 미리 로드해 에이전트의 기능을 확장할 수 있다. Windows, macOS, Linux에서 사용할 수 있는 버전 0.2.1이 공개되었다.

https://joinherd.ai/

#multiagent #ide #aicoding #gstack #parallelprocessing

From the .NET blog...

In case you missed it earlier...

Microsoft Agent Framework – Building Blocks for AI Part 3
https://devblogs.microsoft.com/dotnet/microsoft-agent-framework-building-blocks-for-ai-part-3/ #dotnet #AI #csharp #AIagents #MicrosoftAgentFramework #multiagent #ToolCalling #workflows

From the .NET blog...

Microsoft Agent Framework – Building Blocks for AI Part 3
https://devblogs.microsoft.com/dotnet/microsoft-agent-framework-building-blocks-for-ai-part-3/ #dotnet #AI #csharp #AIagents #MicrosoftAgentFramework #multiagent #ToolCalling #workflows

Ruflo: Multi-agent AI orchestration for Claude Code

https://github.com/ruvnet/ruflo/blob/main/docs/USERGUIDE.md

#HackerNews #Ruflo #MultiAgent #AI #ClaudeCode #Orchestration #GitHub

[Перевод] Как мы перешли на Opus и стали платить меньше

На прошлой неделе мы писали о том, как скармливали терабайты CI-логов LLM . Большинство вопросов на Hacker News касались не самих логов — спрашивали про агента: какие модели, как они взаимодействуют и во сколько всё это обходится. Сейчас мы работаем на Opus 4.6 и платим меньше, чем когда всё крутилось на Sonnet 4.0. Причина в основном в том, чего Opus не делает : 80% сбоев до него не доходят, а когда доходят — он не читает ни одной строки лога. Архитектура выглядит так...

https://habr.com/ru/articles/1030850/

#LLMагенты #multiagent #Claude_Opus #Claude_Haiku #оркестратор #triager #ClickHouse #семантический_поиск #стоимость_инференса

Rohan Paul (@rohanpaul_ai)

Chatly가 Omni Agent를 발표했다. 사용자가 목표를 입력하면 여러 전문 에이전트로 작업을 분산해 처리하는 멀티 에이전트 라우팅 방식으로, 이메일 캠페인, 프레젠테이션, 랜딩 페이지 제작 같은 업무를 자동화한다.

https://x.com/rohanpaul_ai/status/2050641511596109905

#chatly #multiagent #automation #agents #productivity

Title: P2: I was preparing for interview for Agentic AI Engineer. [2025-05-15 Thu]

interpretable outputs (e.g., natural language
explanations) to facilitate human-in-the-loop oversight.

standards:
- FIPA Standards/specifications (e.g., FIPA-ACL)
- IEEE 1471 (ISO/IEC 42010)

I thing, first step of planning multi-agent system is to
measure what intellectual capabilities we can accure
and manage. #dailyreport #multiagent #agents #agentoriented #agentengineering

Title: P1: I was preparing for interview for Agentic AI Engineer. [2025-05-15 Thu]

- Event-Driven Communication: (e.g., Apache Kafka, RabbitMQ)
- Standardized Communication Protocols: like FIPA
(Foundation for Intelligent Physical Agents) for agent
communication, ensuring interoperability and structured message
passing (e.g., ACL - Agent Communication Language).
- Explainability in Communication: Ensure agents provide #dailyreport #multiagent #agents #agentoriented #agentengineering

Title: P0: I was preparing for interview for Agentic AI Engineer. [2025-05-15 Thu]

Main steps in Agent-Oriented Software Engineering:
1) analyzing requirements - what to do and how fast
2) defining system scope and limits - where to run and
how much money we have
3) identifying agent roles
4) selecting architectures
5) choosing tools

best practices:
- microservices REST or gRPC #dailyreport #multiagent #agents #agentoriented #agentengineering

Shreyas Arun (@shreyas_ar54005)

worktree 방식으로 여러 저장소에서 에이전트를 병렬 실행하면, '한 명의 개발자'가 아니라 '한 명의 아키텍트가 팀을 지휘하는 구조'로 바뀐다고 평가합니다. 저장소 간 컨텍스트 공유 방식에 대한 관심도 드러냅니다.

https://x.com/shreyas_ar54005/status/2047925822703669545

#worktree #agents #developerworkflow #git #multiagent