Du wirst angemeldet...

Bitte warte, während wir deine Anmeldung überprüfen

Artikel · Sonntag, 20. September 2026

Agentic Coding

Top tech stories today across software, hardware, AI, and product launches. Senior engineer audience — skip rumour churn and pre-announcement leaks. Lead with shipping-now stories and what changed for builders.

Von Marius BongartsTech104 Ausgaben
← Zur aktuellen Ausgabe
Ausgaben
2 / 104
Über Nacht von KI aus öffentlichen Quellen erstellt, täglich aktualisiert.
Agentic Coding
Sonntag, 20. September 2026
AI Agents - Agentic Coding

APM matures with governance, Coursera ships Cowork course, ARIS adds review loops

1 Min. Lesezeit

APM Governance Features

Your agent config just got enterprise-grade guardrails.

Microsoft's APM now includes security scanning that flags compromised packages before they hit your machine, plus organizational policy enforcement that lets teams whitelist approved sources and primitives [Source: GitHub]. The integrity hashing means your lockfile catches drift between machines—same guarantee you expect from npm or pip. Zero-config Copilot support via apm compile rounds out the tooling. Following up on the coverage from four days ago, this moves APM from promising to production-ready.

Worth revisiting if you skipped the first look.

Coursera Cowork Course

Structured training for Cowork just landed.

A new Coursera course walks you through building full-stack apps with Claude Code, then extends into Cowork agents for workflow automation across Gmail, Slack, and Salesforce via MCP [Source: Coursera]. You'll develop Agent Skills, connect external business tools, and deploy multi-agent systems that handle project management with automated reporting. The curriculum assumes you're past basics and ready for orchestration patterns that span coding and knowledge work.

Fills a gap if you've been piecing together Cowork from docs alone.

ARIS Cross-Model Review

Overnight research just got a verification layer.

ARIS—the markdown-only skill set for autonomous ML research—now chains cross-model review loops into its experiment automation, letting multiple models critique each other's outputs before you wake up [Source: GitHub]. The pattern fits the intent-continuity approach covered earlier this week: structured verification before the next step, not blind chaining. No framework lock-in means you can drop these skills into your existing Claude Code setup.

Test it on your next research spike that can wait until morning.

Quellen
Auto-claude-code-research-in-sleep (ARIS ⚔️ ) - GitHub
Auto-claude-code-research-in-sleep (ARIS ⚔️ ) - GitHub
19 hours ago ... ... agent and the workflows still work. Fork it, rewrite it, adapt it to your stack. Why not self-play with a single model? Using Claude Code subagents or agent ...
github.com
microsoft/apm: Agent Package Manager - GitHub
microsoft/apm: Agent Package Manager - GitHub
17 hours ago ... Why APM. AI coding agents need context to be useful — standards, prompts, skills, plugins — but today every developer sets this ...
github.com
KI-Zusammenfassung

APM (Agent Package Manager) is an open-source dependency manager for AI agents that works with Claude Code, Cursor, GitHub Copilot, and other coding tools. It solves the problem of setting up agent context manually by using a single apm.yml manifest file that declares skills, prompts, plugins, and MCP servers—similar to package.json or requirements.txt. Once declared, `apm install` reproduces the exact same configuration across every developer's machine, with transitive dependency resolution and support for installing from any git host. The tool emphasizes security by scanning for compromised packages and tracking integrity hashes, while governance features let organizations enforce policy on what sources and primitives are allowed. It also supports zero-config setup for Copilot with `apm compile` and integrates with agentrc for generating tailored agent instructions from your codebase.

Quelle öffnen
Claude Cowork, Skills and Plugins for Workflow Automation | Coursera
Claude Cowork, Skills and Plugins for Workflow Automation | Coursera
12 hours ago ... ... Claude Code terminal interface. Design and manage teams of autonomous Claude Cowork agents to automate complex multi-step workflows. Connect Claude to ...
coursera.org
KI-Zusammenfassung

This course covers building full-stack applications with Claude Code and mastering Claude Cowork agents for workflow automation across multiple domains. It includes hands-on training on connecting Claude to external business tools like Gmail, Slack, and Salesforce via Model Context Protocol (MCP), developing Agent Skills, and deploying autonomous multi-agent systems for complex project management with automated reporting and data analysis capabilities.

Quelle öffnen
Über Nacht zusammengestellt von MorningMail.aiZugestellt um 04:55