#agentskills — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #agentskills, aggregated by home.social.
-
Claude Codeにオレたち流のコードを書かせる（中編）— 組織のルールを共有する
https://qiita.com/hiropon122/items/e6b293961d72bd099ede?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Claude Codeにオレたち流のコードを書かせる（中編）— 組織のルールを共有する
https://qiita.com/hiropon122/items/e6b293961d72bd099ede?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Agent-skills-eval – Test whether Agent Skills improve outputs
https://github.com/darkrishabh/agent-skills-eval
#HackerNews #AgentSkills #Eval #Testing #Skills #Outputs #AIResearch #MachineLearning
-
There are many aspects of writing #agentskills that are slightly alien to someone raised on imperative programming. I have to say though being able to just say "spawn one sub-agent per issue in parallel to improve efficiency" has a certain liberating aspect to it.
-
Claude Codeにオレたち流のコードを書かせる（前編）— プロジェクトの流儀を自動抽出する
https://qiita.com/hiropon122/items/80e76424effeeb5431c2?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Claude Codeにオレたち流のコードを書かせる（前編）— プロジェクトの流儀を自動抽出する
https://qiita.com/hiropon122/items/80e76424effeeb5431c2?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Following Andrej Karpathy's example, I've added an LLM "idea file" to the shadow persona adversarial review system for LLM agents that I've been experimenting with for some time. I'm building it as a plugin for Claude, and the idea is to force the agent to refine and incorporate feedback from reviewing personas before responding.
https://github.com/jsynowiec/llm-shadow-persona/blob/main/shadow-persona.md
#llm #generativeai #claude #agentskills #llmIdeaFile #AI #aiagent
-
I've shared two personal Claude Code plugins on GitHub: stackshift and book-forge.
Stackshift helps modernize legacy codebases and reduce technical debt, while book-forge builds ePub e-books from Markdown sources.
I've been actively using both in various projects and documentation workflows for some time now and find them quite helpful.
#Claude #ClaudeCode #AgentSkills #AI #generativeAI #OpenSource
-
What if you could package, distribute, and secure Agent Skills like you do with your container images, configuration, and libraries? This is my proposal for an open specification for managing Skills as OCI Artifacts, re-using the same tools and infra you already have today, and addressing the challenges of the current Git-based approach.
-
----------------
🛠️ Tool — Elastic Agent Skills
===================Overview
Elastic Agent Skills is an official library of packaged instructions and context designed to teach agentic AI tools how to work correctly with Elastic products. The repository targets agentic IDEs and AI coding agents such as Cursor, GitHub Copilot, Windsurf, and Gemini CLI and follows the Agent Skills open standard (agentskills.io). The release is marked as Technical Preview and is under active development.
What the skills contain
Each skill is contained in its own folder and includes a SKILL.md file that provides metadata and agent-facing instructions. Skills codify patterns and context for tasks including interacting with Elasticsearch APIs (search, indexing, cluster management), building and managing Kibana content (alerts, connectors), and applying patterns for Elastic Observability, Elastic Security, and Agent Builder workflows.
Compatibility and scope
Skills are intended to drop into existing agent runtimes so that AI assistants stop using outdated patterns and instead follow curated, updated behavior. Supported agent runtimes listed in the repository include Cursor, GitHub Copilot, Windsurf, and Gemini CLI.
Security considerations emphasized in the repository
The repository explicitly warns that AI coding agents run with real credentials and tool access. It highlights several concrete risks: exposure of security data (PII, embedded credentials) to models, attacker-controlled input that enables prompt injection, and the potential for models to execute dangerous actions when afforded broad permissions. The README references CISA guidance as a baseline for threat modeling and recommends scoping privileges, restricting tool access and network reach, and starting evaluations in isolated, non-production environments.
Limitations and status
Skills are in early release, labeled Technical Preview, and may change as formal evaluations and model evolution occur. The project is open source so teams can audit skills before use. Absent are any runtime or installation instructions; the repository focuses on what skills provide and on guidance for secure evaluation.
🔹 elastic #elasticsearch #agentskills #ai #kibana
🔗 Source: https://github.com/elastic/agent-skills
-
Strands AgentsにAgent Skillsがきたぞ～
https://qiita.com/moritalous/items/5100aae89b9686865810?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Strands AgentsにAgent Skillsがきたぞ～
https://qiita.com/moritalous/items/5100aae89b9686865810?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
via @dotnet : Extend your coding agent with .NET Skills
https://ift.tt/B6GLaeX
#dotnetskills #codingagents #netdevelopment #agentskills #copilotCLI #vscode #dotnetrocks #aiintegration #devblog #dotnet #pluginsmarketplace #codingproductivity #contextualAI #agentp… -
via @dotnet : Extend your coding agent with .NET Skills
https://ift.tt/B6GLaeX
#dotnetskills #codingagents #netdevelopment #agentskills #copilotCLI #vscode #dotnetrocks #aiintegration #devblog #dotnet #pluginsmarketplace #codingproductivity #contextualAI #agentp… -
via @dotnet : Extend your coding agent with .NET Skills
https://ift.tt/B6GLaeX
#dotnetskills #codingagents #netdevelopment #agentskills #copilotCLI #vscode #dotnetrocks #aiintegration #devblog #dotnet #pluginsmarketplace #codingproductivity #contextualAI #agentp… -
via @dotnet : Extend your coding agent with .NET Skills
https://ift.tt/B6GLaeX
#dotnetskills #codingagents #netdevelopment #agentskills #copilotCLI #vscode #dotnetrocks #aiintegration #devblog #dotnet #pluginsmarketplace #codingproductivity #contextualAI #agentp… -
via @dotnet : Extend your coding agent with .NET Skills
https://ift.tt/B6GLaeX
#dotnetskills #codingagents #netdevelopment #agentskills #copilotCLI #vscode #dotnetrocks #aiintegration #devblog #dotnet #pluginsmarketplace #codingproductivity #contextualAI #agentp… -
#Unbrowse — agent-native browser skill for #AI agents. Auto-discovers APIs from browser traffic & generates reusable skills on the fly #opensource #agentskills
🚀 Replaces slow browser automation: captures real network traffic & reverse-engineers API endpoints directly
🧵👇⚡ 100x faster execution (50-200ms vs. seconds) & 80% cheaper — skips DOM waits, selector drift & repeated render/parse loops
-
#Unbrowse — agent-native browser skill for #AI agents. Auto-discovers APIs from browser traffic & generates reusable skills on the fly #opensource #agentskills
🚀 Replaces slow browser automation: captures real network traffic & reverse-engineers API endpoints directly
🧵👇⚡ 100x faster execution (50-200ms vs. seconds) & 80% cheaper — skips DOM waits, selector drift & repeated render/parse loops
-
#Unbrowse — agent-native browser skill for #AI agents. Auto-discovers APIs from browser traffic & generates reusable skills on the fly #opensource #agentskills
🚀 Replaces slow browser automation: captures real network traffic & reverse-engineers API endpoints directly
🧵👇⚡ 100x faster execution (50-200ms vs. seconds) & 80% cheaper — skips DOM waits, selector drift & repeated render/parse loops
-
#Unbrowse — agent-native browser skill for #AI agents. Auto-discovers APIs from browser traffic & generates reusable skills on the fly #opensource #agentskills
🚀 Replaces slow browser automation: captures real network traffic & reverse-engineers API endpoints directly
🧵👇⚡ 100x faster execution (50-200ms vs. seconds) & 80% cheaper — skips DOM waits, selector drift & repeated render/parse loops
-
#Unbrowse — agent-native browser skill for #AI agents. Auto-discovers APIs from browser traffic & generates reusable skills on the fly #opensource #agentskills
🚀 Replaces slow browser automation: captures real network traffic & reverse-engineers API endpoints directly
🧵👇⚡ 100x faster execution (50-200ms vs. seconds) & 80% cheaper — skips DOM waits, selector drift & repeated render/parse loops
-
via #AIFoundry : Building an AI Skills Executor in .NET: Bringing Anthropic’s Agent Pattern to the Microsoft Ecosystem
https://ift.tt/dj92LK6
#AI #Anthropic #AgentSkills #ModelContextProtocol #MCP #MCPServer #DotNet #NET8 #CSharp #AzureOpenAI #OpenAI #LLM #EnterpriseAI #Softw… -
via #AIFoundry : Building an AI Skills Executor in .NET: Bringing Anthropic’s Agent Pattern to the Microsoft Ecosystem
https://ift.tt/dj92LK6
#AI #Anthropic #AgentSkills #ModelContextProtocol #MCP #MCPServer #DotNet #NET8 #CSharp #AzureOpenAI #OpenAI #LLM #EnterpriseAI #Softw… -
via #AIFoundry : Building an AI Skills Executor in .NET: Bringing Anthropic’s Agent Pattern to the Microsoft Ecosystem
https://ift.tt/dj92LK6
#AI #Anthropic #AgentSkills #ModelContextProtocol #MCP #MCPServer #DotNet #NET8 #CSharp #AzureOpenAI #OpenAI #LLM #EnterpriseAI #Softw… -
via #AIFoundry : Building an AI Skills Executor in .NET: Bringing Anthropic’s Agent Pattern to the Microsoft Ecosystem
https://ift.tt/dj92LK6
#AI #Anthropic #AgentSkills #ModelContextProtocol #MCP #MCPServer #DotNet #NET8 #CSharp #AzureOpenAI #OpenAI #LLM #EnterpriseAI #Softw… -
via #AIFoundry : Building an AI Skills Executor in .NET: Bringing Anthropic’s Agent Pattern to the Microsoft Ecosystem
https://ift.tt/dj92LK6
#AI #Anthropic #AgentSkills #ModelContextProtocol #MCP #MCPServer #DotNet #NET8 #CSharp #AzureOpenAI #OpenAI #LLM #EnterpriseAI #Softw… -
Claude Skills: Office-Dokumente im eigenen Corporate Design automatisch erstellen
→ https://internet-fuer-architekten.de/claude-skills-office-dokumente-corporate-design/#AgentSkills #ClaudeSkills #CorporateDesign #KünstlicheIntelligenz
-
Skill.md: An open standard for agent skills
https://www.mintlify.com/blog/skill-md
#HackerNews #SkillMD #OpenStandard #AgentSkills #AIDevelopment #TechInnovation
-
Skill.md: An open standard for agent skills
https://www.mintlify.com/blog/skill-md
#HackerNews #SkillMD #OpenStandard #AgentSkills #AIDevelopment #TechInnovation
-
Using Agent Skills to add alt texts to my blog post images
https://www.jannemattila.com/appdev/2026/01/12/using-agent-skills.html
#GitHubCopilot #AgentSkills -
RE: https://hachyderm.io/@kejne/115867659083244056
DIdn't realize this repository is only about a week old and already got 4k ⭐ !
-
I recently organized moved data which I had stored on Google Drive in documents and sheets into plain markdown and restructuring the data into linked notes, using #obsidian.
For example recipe links that pointed to different sites as well as recipes in a giant document.
I asked #codex to import one sample recipe for me to validate and then it could get going and nicely structure the notes together with relevant links and tags (I wanted the recipes tagged with the protein group and type of fish).
I think this is an excellent example of where #agentskills can provide an immediate useful boost to the result.
Cool to see repos with specialized skills popping up. This proved very useful for my vault!
https://github.com/kepano/obsidian-skills -
Google DriveのドキュメントをAIエージェント（Claude Code等）のナレッジにするシステムを作ってみた
https://qiita.com/hiropon122/items/e29c2f989c2c42b4b6a0?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Google DriveのドキュメントをAIエージェント（Claude Code等）のナレッジにするシステムを作ってみた
https://qiita.com/hiropon122/items/e29c2f989c2c42b4b6a0?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Anthropic, AI 에이전트 스킬 오픈 스탠다드로 공개 – OpenAI도 따라한 기술
Anthropic이 Agent Skills를 오픈 스탠다드로 공개하며 AI 에이전트 표준화를 주도합니다. OpenAI도 따라한 이 기술의 전략적 의미를 분석합니다. -
#Anthropic launched its #AgentSkills technology as an #openstandard, aiming to establish itself in the #enterprisesoftware market. The technology, which enables #AIassistants to perform specialised tasks, is now being adopted by companies like Microsoft and OpenAI. https://venturebeat.com/ai/anthropic-launches-enterprise-agent-skills-and-opens-the-standard?AIagents.at #AIagent #AI #ML #NLP #LLM #GenAI
-
Anthropic just unveiled Agent Skills, a new toolkit that lets Claude act like a plug‑in‑ready assistant across GitHub, Atlassian, Figma and more. Early adopters report impressive results, and the move directly challenges OpenAI’s roadmap. Curious how it works? Read on! #Anthropic #AgentSkills #OpenAI #Claude
🔗 https://aidailypost.com/news/anthropic-launches-agent-skills-challenges-openai-sees-strong-feedback
-
👨💻 Oh great, another "revolutionary" AI tool because clearly, what we all needed was a way to copy and paste the same tired "Agent Skills" across Claude's various platforms. 🚀 It's the future of AI, folks—where you can dazzle your colleagues with your cutting-edge ability to master toggle modes and governance instructions. 🙄
https://claudeskills.cc #AItools #AIrevolution #AgentSkills #TechTrends #Innovation #HackerNews #ngated