QuACS18 stories · last 7 days · 5 newsletters + 3 web sources
Vibe & agentic coding
Claude Code Now Manages Multiple Agents in a Single Terminal Window
Claude Code has shipped a new feature allowing users to view and manage all running AI coding agents in one terminal window, with inline status updates and the ability to unblock agents needing input via /bg. Anthropic has also added a multi-agent dashboard to Claude Code, enabling better visibility and management of multi-agent workflows directly within the tool.
█████ Ben’s Bites, The Neuron
Screen-recording as agent feedback: video-to-HTML workflow skill
A practical agentic coding workflow where you screen-record your feedback (including showing reference sites), pass the video file to your agent, and it generates a structured HTML report with GIFs and an actions checklist. The author shares a reusable agent skill prompt (video-to-html) you can drop into your own workflow immediately.
█████ Ben’s Bites
Claude Plan Changes for Third-Party Tool Usage Starting June 15th
Anthropic is introducing a separate usage limit for Claude accessed via third-party tools (e.g. Cursor, Zed, T3 Code, Conductor), providing extra monthly credits equal to your plan value but with no rollover and API-rate billing after exhaustion. Claude Code and the Claude app remain on their existing separate limits, so users of agentic coding tools need to review their usage before June 15th.
████░ Ben’s Bites, The Rundown AI
OpenCode: Free Open-Source Claude Code Alternative
OpenCode is an open-source alternative to Claude Code that is gaining traction among developers. This is directly relevant to anyone using or evaluating agentic coding tools like Claude Code, Cursor, or Copilot.
████░ The Neuron
OpenAI’s Codex Moves to Mobile for Remote Agentic Coding Management
OpenAI has rolled out Codex in preview inside the ChatGPT iOS app, allowing developers to monitor, approve, and manage long-running AI coding agent tasks from their phone while the agent continues working on a desktop or remote host. This directly extends agentic coding workflows beyond the desk, enabling async oversight of autonomous coding pipelines.
████░ The Rundown AI
Agentic Coding Is a Trap — Learn the System, Not the Syntax
A blog post argues that true agentic coding requires understanding underlying systems rather than just prompting AI with syntax, distinguishing ‘vibe coding’ from ‘agentic engineering’. Directly relevant for anyone building with AI coding tools, as it reframes how to approach skill-building and workflow design.
████░ Ben’s Bites
I Don’t Think AI Will Make Your Processes Go Faster
AI cannot resolve bottlenecks caused by vague requirements and incomplete documentation — the real source of delays in software development workflows. A useful reality check for teams building agentic coding or automation pipelines who may be over-indexing on AI execution speed.
███░░ TLDR AI
We stopped AI bot spam in our GitHub repo using Git’s –author flag
A team shares how they identified and blocked AI-generated bot spam in their GitHub repository using Git’s built-in author filtering. Useful for teams managing agentic coding workflows where automated agents commit code and repo hygiene becomes a concern.
███░░ Hacker News
AI agents & automation
Solo Operator Builds 7-Agent Claude Code System for B2B Cold Email Automation
A viral post describes a multi-agent Claude Code setup with one orchestrator and six sub-agents using MCP servers to run cold email campaigns autonomously, with humans only approving exceptions. The workflow pattern — specialized sub-agents, quality evals, and exception-based human oversight — is a directly actionable blueprint for agentic coding and automation workflows.
████░ The Neuron
OpenAI Codex to Control Desktop Devices via Computer Use While Locked/Asleep
OpenAI is developing a capability for its Codex coding agent to operate macOS applications through Computer Use even when the machine is locked or asleep, removing the need for an active session. This directly expands agentic coding workflows by enabling truly autonomous, unattended operation of desktop dev environments.
████░ TLDR AI
Google DeepMind’s AI Co-Mathematician Uses Agentic Workflow Modeled on Claude Code
DeepMind built an agentic math research system where a coordinator agent breaks problems into parallel workstreams with sub-agents handling code, literature search, and proofs — explicitly modeled after AI coding environments like Claude Code. This is a direct, concrete example of multi-agent orchestration architecture that mirrors agentic coding workflows, offering transferable design patterns for agent builders.
████░ The Rundown AI
Claude’s Next Enterprise Battle Is Not Models: It’s the Agent Control Plane
Microsoft leads enterprise agent orchestration with 38.6% adoption as enterprises shift focus to control planes for managing agent workflows, permissions, and auditability. This matters for anyone building or deploying agentic systems, as governance and security are becoming the key differentiators over raw model capability.
████░ TLDR AI
Custom MCP Catalogs and Profiles: Advancing Enterprise MCP Adoption
Docker has launched Custom MCP Catalogs and Profiles, allowing enterprises to distribute curated, trusted MCP server collections and portable configurations for AI agents. This directly improves agentic coding and multi-agent workflows by reducing context window bloat and standardizing tool access across teams.
████░ TLDR AI
Context Pruning: Cut LLM Tokens Without Losing Quality
Context pruning removes low-value tokens or passages from LLM inputs, achieving up to 20x compression with faster latency and maintained output quality. Directly useful for agentic coding workflows and AI pipelines where long context windows and token costs are a practical concern.
████░ TLDR AI
Show HN: InsForge – Open-source Heroku for coding agents
InsForge is an open-source platform designed to deploy and host coding agents, positioning itself as a Heroku-style infrastructure layer for agentic coding workflows. Directly relevant to anyone building or deploying autonomous coding agents and agentic pipelines.
████░ Hacker News
Thinking Machines Lab Introduces ‘Interaction Models’ for Real-Time AI Collaboration
Mira Murati’s TML unveiled interaction models that process voice, video, and text in 200ms streaming chunks, with a separate background model handling reasoning and tool use — positioning it as an alternative to purely agentic AI architectures. This dual-model design (live interaction + background agent) is directly relevant to understanding new patterns in agentic workflow orchestration and multi-agent system design.
███░░ The Rundown AI
QA & testing
Inngest Report: 69% of Engineers Building AI Agents in Production Cite Reliability and Tracing Gaps
A survey of 130 engineers found only 19% were confident their AI agent stack could scale, with tracing gaps and reliability work consuming up to half of some teams’ time. This is directly actionable for anyone building or evaluating agentic workflows and QA pipelines in production.
████░ TLDR AI
Defense at AI Speed: Microsoft’s New Multi-Model Agentic Security System (MDASH)
Microsoft’s MDASH autonomously scans for and fixes software vulnerabilities, achieving 96–100% recall on real-world Windows bugs — representing a shift from AI-assisted to fully autonomous code remediation. This is directly relevant as an example of agentic workflows applied to automated code QA and vulnerability testing.
███░░ TLDR AI
Sources
Newsletters: The Neuron, The Rundown AI, TLDR AI, Ben’s Bites, Import AI
Web: TechCrunch AI, Hacker News, Simon Willison
Generated by ai-digest-cli on 2026-05-18 16:44