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.
← Zur aktuellen AusgabePersistent agent memory ships, coding-agent ecosystem explodes past 4,800 repos
AgentMemory Persistent Context
Your agents can finally remember yesterday's session.
AgentMemory is a new persistent memory system that captures every tool use through lifecycle hooks, compresses observations into searchable memory, and injects relevant context when new sessions start [Source: GitHub]. It works across 20+ agents including Claude Code, Cursor, Codex CLI, and Gemini CLI through native plugins, MCP servers, or REST APIs. The hybrid search combines keyword matching with semantic vectors and knowledge graphs across four memory tiers—working, episodic, semantic, and procedural. Token costs drop from 650K to 170K annually compared to LLM-summarized context, or near-zero if you run local embeddings.
Pairs perfectly with your AGENTS.md setup for cross-session continuity.
Coding Agent Repo Count
The coding-agent ecosystem just hit a milestone you can browse.
GitHub's coding-agents topic page now lists 4,826 public repositories—up dramatically from the 130+ CLI agent catalog covered yesterday [Source: GitHub]. Python leads with 1,456 repos, TypeScript follows with over 1,100. You'll find everything from orchestration frameworks to specialized context compression tools. The topic page works as a discovery layer when you're hunting for niche capabilities the curated lists haven't caught yet.
Worth filtering by language and stars before your next tool audit.
Agent Workflow Mindset
The bottleneck isn't your tooling—it's how you describe what you want.
A growing thread in the agentic coding community argues that people who thrive with agents share one trait: they can articulate exactly what they need before touching the keyboard [Source: Facebook]. The consensus is that vague prompts produce vague output regardless of model quality. Structured PRDs and explicit acceptance criteria outperform improvisation every time. This reinforces yesterday's dependency-chain approach—precision beats volume.
Your next Claude Code session might start with ten minutes of writing, not coding.
rohitg00/agentmemory: #1 Persistent memory for AI coding agents ...13 hours ago ... ... skills add rohitg00/agentmemory -y # 17 native skills so your agent knows when to reach for memory. The keyword searches should hit in default keyless mode ...github.com
Mindset is key to succeeding with AI and vibe coding - Facebook14 hours ago ... And side note: I DID vibe code a destination description builder along with a packing list that my assistant uses to help flesh out itineraries. 45m. Gina Matt ...facebook.com
coding-agents · GitHub Topics24 hours ago ... ... prompt-engineering anthropic coding-agents agentic-ai ai-coding claude-code loop-engineering. Updated in 13 hours; TypeScript. EKKOLearnAI / ekko-studio · Star ...github.com