VS-CODE PUB_DATE: 2026.05.22

VS CODE BRINGS AUTO MODE AND REMOTE AGENT SESSIONS TO CLAUDE; CLAUDE CODE REVAMPS /CODE-REVIEW

Visual Studio Code now lets the Claude Agent run remotely and, in preview, act without constant prompts, while Claude Code changes how reviews work. Microsoft ...

Visual Studio Code now lets the Claude Agent run remotely and, in preview, act without constant prompts, while Claude Code changes how reviews work.

Microsoft shipped VS Code 1.121 with preview support for running agent sessions on remote machines over SSH/dev tunnels and an Auto mode for the Claude Agent that executes without per‑step prompts, gated by a safety classifier Visual Studio Magazine. This pushes AI agents closer to unattended workflows inside the editor.

In parallel, Claude Code v2.1.147 replaced /simplify with /code-review that reports correctness bugs at selectable effort levels and can post inline GitHub PR comments with --comment; it also made pinned background sessions more durable. v2.1.148 quickly fixed a Bash exit code regression (v2.1.147, releases).

[ WHY_IT_MATTERS ]
01.

Auto mode and remote sessions shift Claude from chatty assistant to semi-unattended worker, changing review, security, and ops policies.

02.

Claude Code’s new /code-review can become a CI gate that flags correctness issues and comments directly on PRs.

[ WHAT_TO_TEST ]
  • terminal

    Spin up a VS Code remote agent session over SSH against a disposable repo; measure tool access, latency, and audit logging coverage.

  • terminal

    Enable Claude Agent Auto mode in a sandbox branch and compare defect catch rate and misfire rate versus prompt-permission mode using seeded tasks.

[ BROWNFIELD_PERSPECTIVE ]

Legacy codebase integration strategies...

  • 01.

    Pilot /code-review in CI on a subset of services; require --comment to annotate PRs and track signal-to-noise and false positives.

  • 02.

    Define guardrails before turning on Auto mode: allowed commands, repo scope, secrets redaction, and rollback paths; verify logs meet compliance.

[ GREENFIELD_PERSPECTIVE ]

Fresh architecture paradigms...

  • 01.

    Design repos for agent execution from day one: least-privilege service accounts, ephemeral runners, and clear test oracles agents can validate against.

  • 02.

    Adopt agent-first workflows: routines to run tests, apply migrations, and open PRs automatically on schedule or on pipeline events.

Enjoying_this_story?

Get daily VS-CODE + SDLC updates.

  • Practical tactics you can ship tomorrow
  • Tooling, workflows, and architecture notes
  • One short email each weekday

FREE_FOREVER. TERMINATE_ANYTIME. View an example issue.

GET_DAILY_EMAIL
AI + SDLC // 5 MIN DAILY