September 10, 2026: Smart reports (beta) for Enterprise
Claude can assemble smart reports on Enterprise plans, in beta.
What this means for your agents: Recurring reporting agents (pipeline, cash flow, content) get a native report surface instead of a pasted chat answer.
September 1, 2026: Claude Fable 5.1 and Mythos 5.1
Two new Claude models shipped: Fable 5.1 and Mythos 5.1.
What this means for your agents: Fresh model options for agent runs - worth re-testing your agent’s prompts against the new models before switching the default.
August 25, 2026: Memory across chat and Cowork, with editable topics
Claude’s memory now works across chat and Cowork, and you can edit what it remembers by topic.
What this means for your agents: Your agent stops re-asking setup questions, and you can prune wrong memories directly instead of prompt-patching around them.
August 6, 2026: Security scanning for skills and plugins (beta)
Claude added beta security scanning for third-party skills and plugins.
What this means for your agents: A real safety layer for exactly what this store sells - third-party skill files get scanned before they run.
July 24, 2026: Claude Opus 5
Anthropic released Claude Opus 5, the new top-tier model.
What this means for your agents: The ceiling for hard agent tasks (long reconciliations, multi-document diligence) moved up.
More Claude agent resources
AI agent skills for Claude - the honest setup guide for running our 28 agents on Claude. Start with the Sourcing Agent on Claude guide if you want a concrete example.
AI Agent Skills is an independent store. We are not affiliated with, endorsed by, or sponsored by Claude or its makers. Release details belong to their sources; check Claude's own documentation for the authoritative record.