Coding agents & CLI tooling

• Flue v2: Astro creator's meta-harness ships React-style "Agent Hooks" as its first stable release, another entrant in the emerging agent-framework space alongside Vercel's eve.
• Superagent: pitched as "Claude Code for the rest of us", another consumer-friendly wrapper around agentic coding.
• Forge: new Claude Code plugin targeting agentic dev workflows.
• Claude CLI subagents: community thread on subagent patterns in Claude Code.
• awesome-mcp-servers: curated MCP server list, useful reference when wiring tool integrations.
• oMLX: Mac-local LLM server claiming to cut agent wait times from 90s to 5s, relevant if running local models for coding agents.

LLM mechanics & context

• Opus 4.6 1M context in Claude Code, still available?: community check-in on long-context availability, worth tracking for context-window budgeting.
• Understanding ChatGPT Work: Simon Willison unpacks two products under one name; shared Codex quota reportedly makes the cloud version impractical for power users despite strong background-task capability.
• Opus 4.6 randomly "not thinking" in Claude Code Desktop: reported cause of extended-thinking silently dropping mid-session.
• MEM-ABBREV: citations, memory compression, sycophancy resistance: community technique for compressing Claude's memory/context footprint.
• Removing Claude's default user-profile context window: trick to reclaim context budget Claude auto-allocates to a user profile.
• Continuous Diffusion Language Models: survey of a resurgence in continuous (non-autoregressive) diffusion approaches for language, an alternative generation paradigm to today's LLMs.
• How to build a diffusion language model: ICLR/MLSS 2026 material walking through masking diffusion, iterative refinement, and variable-length generation for open-source diffusion LLMs.
• What five Windows apps expose to a computer-use agent: measured attack/access surface before granting computer-use agents app control, relevant for agent security posture.

Vibe-coding & repos

• crawl4ai: open-source LLM-friendly web crawler turning pages into clean Markdown for RAG/agent pipelines, now positioning as cost-effective at scale.
• PRAXIST: autonomous research system coordinating parallel research peers with durable evidence, treats research as a persistent process rather than one-off prompts.
• last30days-skill: Claude Code skill that researches a topic across Reddit/X/YouTube/HN/Polymarket and synthesizes a grounded summary scored by engagement, not editors.

Agentic SaaS

• Maritime: dedicated $1/month compute instances for AI agents.
• Referent: AI-native OS pitched for law firms.
• Olostep: web-to-clean-data API for feeding AI pipelines.
• reverse-skill: AI-routed skill pack for reverse engineering / authorized pentesting workflows across Claude Code, Kiro, Cursor, Cline.
• Memory layer for Claude and Cursor: third-party persistent-memory add-on so coding agents stop losing context between sessions.

CLI/TUI

• ipatool: CLI to search and download iOS/iPadOS/tvOS/visionOS App Store .ipa packages, cross-platform.