Fetching from the wire…
Tools2026-08-09 · source-backed
The companion post reports that across all Claude Code usage, sessions now run 9x longer between interruptions than under the previous default, and at Gusto roughly 10% of session transcripts contained at least one auto-mode denial. The classifier fires without stalling legitimate work. The recommended pattern is layered, not classifier-only: Gusto routes MCP traffic through a governed proxy with tool guards and prompt inspection before auto mode sees anything, Nuro disables recursive deletes in settings, Garner Health blocks Slack and email sending on the user's behalf. Teams still flip to interactive for production infrastructure: Terraform, AWS, direct POSTs against live APIs. Gusto's Evan Magnussen names the precondition: without telemetry, "that would be very dangerous."
Each link below shares sources, entities, or timing with this story.
Three separate Anthropic changes over about two weeks point the same direction, and none of them announced themselves as a strategy. Claude Code 2.1.238 added claude self-hosted-runner --defer-shutdown-max-min, which keeps serving attached sessions on SIGTERM, parks whatever's...
A spec is a press release until someone who didn't write it implements it. GitHub made Agent Plugins 1.0 generally available on August 12 across VS Code, Copilot CLI, the Copilot SDK, and the Copilot app on all plans. The spec, published August 6, was co-authored by AWS, Anysp...
Beta for Claude Enterprise as of August 11: server-hosted transcripts covering prompts and responses, web and MCP tool-call content, skills and artifacts content captured as transcript text, plus verified user ID and email, org ID, session and per-message IDs, timestamps (Anth...
MCP now has 97M+ monthly SDK downloads. First-class client support across ChatGPT, Claude, Cursor, Gemini, Microsoft Copilot, and VS Code. Governed by the Agentic AI Foundation (AAIF) under Linux Foundation, co-founded by Anthropic, Block, and OpenAI. AAIF also hosts Goose (op...
Launched in beta August 5 for Claude Enterprise, routing each prompt and tool-call response over a signed WebSocket to the customer's own security server. Covers Claude chat, Claude Code, Claude Cowork, and tool calls through MCP connectors, skills and plugins, with named inte...
Output tokens cost roughly 5x input, because decode is sequential. Cache reads cost 0.1x input price. The prompt cache expires after 1 hour on subscriptions and 5 minutes on API keys. Those three numbers are the whole post, and Anthropic put them in one place for the first tim...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.