Five open-source supply chain attacks in twelve days. Axios npm compromised with 100M weekly downloads. TeamPCP's lateral credential chain from Trivy to Checkmarx to LiteLLM to Telnyx to Axios. Copilot rewrote a developer's PR. The Cognitive Dark Forest as signal theory.
Why memory search and context injection are different tools, and how cascading token budgets solve the prompt enrichment problem.
GitHub's private repo training opt-out deadline, three supply chain attacks, Claude Code 2.1.86, ARC-AGI-3's sub-1% AI score, and Ecorobotix's $50M precision spraying investment.
How a research paper on Recursive Language Models became a Rust CLI tool and a Claude Code swarm plugin that processed 75,000+ rows of messy enterprise data and discovered an 11-dimension taxonomy nobody planned.
Astral joins OpenAI in the week's biggest developer tools acquisition. Claude Code ships four releases with MCP push channels and 128k context. NVIDIA opens RL training infrastructure for multi-turn agents.
Claude Code shipped 3 releases in 3 days with a 12x prompt caching fix. Agricultural data ownership enters a new phase. LLM distillation research gets more principled. OpenAPI Overlay v1.1 lands.
GPT-5.4 ships with a system card, a GitHub issue title exploited Cline to compromise 4,000 developer machines, and Claude Code 2.1.70 brings native MCP management to VS Code.
Claude Code Remote Control ships, tool choices get scrutinized, Block cuts half its workforce for AI, MCP hits production, and agent reliability research picks up.
Anthropic raises $30B at a $380B valuation while Gemini 3.1 Pro lands on HN's front page with 806 upvotes. MCP servers from AWS, Microsoft, and GitHub signal the protocol has graduated from experimental to infrastructure.
Two open specs tackle AI memory and extension fragmentation: MIF for memory interchange and ccpkg for packaging. MCP ecosystem hits 700+ servers.