Anthropic / Claude ecosystem
Claude Code v2.1.193 released with MCP and auto-mode improvements
Anthropic released Claude Code v2.1.193 on June 25, introducing incremental improvements to Model Context Protocol (MCP) authentication, auto-mode command classification, and shell command handling. The update adds automatic memory-pressure reaping for background shell commands, improved MCP error messaging with HTTP details, and expanded auto-mode denial reasons in transcripts. Background agents can now continue working while the main session manages other tasks.
- Source: GitHub (Anthropic)
- Significance: routine
Claude Fable 5 returns to service with government security approval
Following a three-day suspension triggered by alleged jailbreak concerns, Claude Fable 5 is returning to service with enhanced security measures in place as of June 26, 2026. The model now requires stricter authentication including government-issued ID verification. Anthropic remains in negotiation with the Trump administration over the more advanced Mythos 5 model, with CEO Dario Amodei stepping back from talks in favor of co-founder Tom Brown. The return signals that US government negotiations over frontier AI safety are resolving.
- Source: Geeky Gadgets
- Significance: moderate
- Update: Prior suspension announced June 12, 2026. Model reintroduction with government security approval signaled June 26, 2026, establishing new identity-gated access requirements for enterprise deployments.