Claude Report — 2026-05-09
- Anthropic signs SpaceX Colossus 1 deal (300MW, 220k GPUs) and doubles Claude Code five-hour rate limits for Pro/Max/Team/Enterprise, dropping peak-hour caps.
- Claude Code 2.1.136 ships `autoMode.hard_deny`, PostToolUse hooks for all tools, MCP/plugin persistence after `/clear`, and enterprise OTEL feedback survey support.
- AWS launches Agent Toolkit for AWS: 40+ official skills, IAM-governed MCP server with CloudTrail audit and approval gates, native Claude Code plugin.
- `claude-api` Agent Skill gains Managed Agents docs (outcomes, multiagent sessions, webhooks) in sync with the May 6 API public beta.
- Claude Code 2.1.137–2.1.138 patch VS Code Windows activation and ship internal stability fixes.
🚀 Anthropic Official
Higher usage limits for Claude and a compute deal with SpaceX — May 6
Anthropic signs a deal for SpaceX's Colossus 1 data center (300MW, 220k NVIDIA GPUs available within one month) and simultaneously doubles Claude Code's five-hour rate limits across Pro, Max, Team, and Enterprise plans while removing peak-hour reductions for Pro and Max subscribers; the announcement also references future exploration of orbital AI compute with SpaceX.
Claude Code 2.1.136–2.1.138 — May 9
2.1.136 adds settings.autoMode.hard_deny for unconditional blocking in auto mode regardless of allow-exceptions, CLAUDE_CODE_ENABLE_FEEDBACK_SURVEY_FOR_OTEL for enterprise session-quality telemetry, a type-to-filter search box in /skills, hookSpecificOutput.updatedToolOutput in PostToolUse hooks for all tools (previously MCP-only), and /mcp now shows tool counts per connected server and flags zero-tool connections; also fixes MCP servers and plugins silently disappearing after /clear in VS Code, JetBrains, and Agent SDK, rare login loops from concurrent credential writes, and OAuth refresh token loss under concurrent server refreshes. 2.1.137 addresses VS Code Windows extension activation failing; 2.1.138 is an internal maintenance build.
🔌 Claude Code Plugins
None of the top 10 updated in window.
Unchanged in window: frontend-design, superpowers, context7, code-review, code-simplifier, github, skill-creator, playwright, feature-dev, claude-md-management.
🛠️ Skills
None of the top 5 updated in window.
Unchanged in window: superpowers:brainstorming, superpowers:tdd, superpowers:systematic-debugging, superpowers:writing-plans, superpowers:executing-plans.
🤖 Agents & Subagents
None of the top 5 updated in window.
Unchanged in window: superpowers:subagent-driven-development, claude-router:orchestrate, ralph-loop, claude-router:fast-executor, claude-mem:do.
🔗 MCPs & Integrations
None of the top 5 updated in window.
Unchanged in window: github-mcp-server, modelcontextprotocol/servers, filesystem MCP, claude-mem, context7.
💡 Community — Workflows & Ideas
AWS Agent Toolkit for AWS — May 6
AWS releases an official bundle of 40+ skills, MCP servers, and a Claude Code plugin at aws/agent-toolkit-for-aws providing IAM-governed access to AWS services with CloudTrail audit of every agent call, per-action approval gates, and read-only mode as the safe default; the Claude Code plugin enters the official marketplace via PR #1756 on May 7.
claude-api skill — Managed Agents outcomes, multiagent, webhooks — May 6
Adds code examples and usage documentation for Managed Agents multiagent sessions, outcomes, and webhook subscription directly into the claude-api Agent Skill, keeping it synchronized with the same-day API public beta.
📰 Quick Mentions
- HN: "Higher usage limits for Claude and a compute deal with SpaceX" — Community flags Colossus 1's environmental-permit history, questions orbital compute economics, and notes potential revenue-inflation optics ahead of Anthropic's IPO.
- Marketplace cleanup May 7 — autofix-bot (#1047) and adspirer-ads-agent (#1716) removed (beyond the three reported yesterday); zapier SHA bumped to f34a785; GitHub Actions pinned to commit SHAs for security hardening.