minssam.
Published on

Claude Code 2.1.310: Monitor Comes to Bedrock, Vertex & Foundry β€” Plus Safety Telemetry and /import

Claude Code 2.1.310 shipped on October 23, 2026. The headline changes point squarely at teams running Claude Code in enterprise and cloud environments, not just individual developers.

Monitor Tool Now Works on Bedrock, Vertex, and Foundry

The Monitor tool lets you watch a background agent or long-running task in real time β€” tracking which tools it called, what output it produced, and where it currently stands. Before this update, Monitor worked reliably only on the standard Claude API. For teams routing requests through AWS Bedrock, Google Vertex AI, or Azure AI Foundry, Monitor was either partially broken or completely unavailable.

2.1.310 fixes that. All three platforms now have full Monitor support, including in environments where telemetry is disabled. The long foreground sleep block β€” which prevents agents from sitting idle too long β€” also now enforces consistently across all three.

For any organization using Bedrock or Vertex due to enterprise contracts, data residency requirements, or compliance policies, this means Claude Code''s observability features are finally available without compromise.

Safety Telemetry: A New Event for Every Blocked Tool Call

The new claude_code.safety_event OpenTelemetry event fires automatically whenever:

  • A permission rule blocks a tool call
  • Auto mode stops execution on a safety check
  • A hook prevents a tool from running
  • The sandbox denies a command

Why does this matter? Until now, understanding why an agent stopped β€” what rule triggered, which tool was blocked, how often it happened β€” required log diving or debug mode. Safety decisions were largely invisible after the fact.

With claude_code.safety_event, you can query: "In the past month, how many times did our agents hit permission blocks, and which tools caused them most often?" That''s data you can actually use to tune safety policies, not just react to incidents after they happen.

/import: Bring Over Settings from Other AI Tools

The /import command is the most practically useful addition for individual developers:

/import codex    β€” import settings from OpenAI Codex CLI
/import gemini   β€” import settings from Gemini CLI
/import cursor   β€” import settings from Cursor

If your team is migrating to Claude Code or running it alongside another tool, there''s no longer a reason to rebuild your configuration from scratch. One command and your existing setup carries over.

This fits a broader pattern that''s been developing in 2026: Claude Code, Codex CLI, and Gemini CLI have converged on shared commands like /clear, /model, and /plan. The /import command is the next logical step β€” making the tools genuinely interoperable instead of isolated silos.

Role-Based Cloud Session Permissions

Enterprise admins get a new control: you can now assign cloud session and Remote Control access by role. Lock down cloud sessions for contractors, open them up for senior engineers. It''s the kind of granular access control that makes Claude Code deployable in larger organizations without blanket permissions.

Setup scripts can now also be loaded from a file rather than entered inline β€” making it much easier to standardize the initial environment across a team.

The Pattern Here

Every change in 2.1.310 β€” Monitor on enterprise clouds, Safety telemetry, /import, role-based access β€” is about making Claude Code work better at scale, inside organizations, not just on a solo developer''s machine. That''s the direction Anthropic seems to be pushing Claude Code: from a sharp tool for individual use toward infrastructure that teams can actually govern and operate.

For individual developers, /import alone is worth the update. For teams on Bedrock or Vertex, the Monitor fix is the real unlock.


Sources:

Claude Code 2.1.310: Monitor Comes to Bedrock, Vertex & Foundry β€” Plus Safety Telemetry and /import | MINSSAM.COM