VS Code 1.137 Adds Scheduled Agent Automations
Software / news
VS Code 1.137 Adds Scheduled Agent Automations
The Preview feature runs a Copilot agent task hourly, daily or weekly without you starting it, while a new protocol lets separate windows share one agent session.
Visual Studio Code 1.137, released Sept. 9, adds a Preview feature called Automations that runs a Copilot agent task on a schedule instead of waiting for a developer to start it, according to Microsoft's own release notes.
Automations runs the task before you open the laptop
Turned on with the chat.automations.enabled setting, Automations lets a developer pick a template built for catching up on changes, triaging issues or finding bugs, or write a custom prompt, then set it to run hourly, daily, weekly or on demand from the Agents window sidebar. Microsoft's release notes describe the feature as rolling out gradually to all users but do not say what share of users have access as of this release, or whether a run counts against the same usage limits as a chat session started by hand.
Voice Mode lets you interrupt an agent mid-sentence
A second, Experimental feature, Voice Mode, adds spoken conversation with an agent through three settings: agents.voice.enabled, agents.voice.showTranscript and agents.voice.voice. A developer can push to talk, interrupt or redirect an agent while it is still responding, and the interface tracks the active session, selected model and attached files as it listens. Organization administrators can turn the feature off by disabling Copilot preview features for their org, per the release notes.
NT Compatible writer Philipp Esselbach, who covered the release the same day, said Voice Mode carries "background overhead" that offsets its time savings on long refactoring sessions, and that both headline features target "power users already running agents regularly" rather than people opening VS Code for the first time.
| Feature | Status | Setting |
|---|---|---|
| Automations | Preview | chat.automations.enabled |
| Voice Mode | Experimental | agents.voice.enabled |
| GitHub issue and PR viewing in Agents window | Experimental | not specified |
The feature neither outlet led with
Esselbach argued the real story in 1.137 is smaller and less demo-friendly: the Agent Host Protocol, which lets multiple VS Code windows connect to the same running agent session instead of each window starting its own. He said that architecture change, decoupling an agent session from any single workspace, is what makes "autonomous agents actually practical instead of just performative," a claim Microsoft's own release notes support with a feature description but do not quantify with any measurement of session continuity or dropped-session rates.
The release also adds a unified diff-editor layout shared between single-file and multi-file review, and live status updates on GitHub issue and pull-request links pasted into Markdown, both listed as Experimental. None of the three Experimental features has a graduation date to General Availability in the release notes. Set against GitHub's own disclosure of Copilot's inference-cost cuts earlier in the week, the pattern in Microsoft's developer tools this week is the same one twice: ship the agent feature first, in preview, and publish the cost and reliability numbers separately, later, or not at all. Understanding what Automations and Voice Mode actually are requires the vocabulary GitHub has been formalizing for its own agent products, catalogued in The Terminal's look at GitHub's AI coding glossary.
Sources
More in Software
- 01Rivet, a YC W23 Company, Runs Agent Actors on S3 and Claims 72 KB EachThe Apache 2.0 orchestrator says an actor cold-starts in 4.99 ms at p99. Every figure is Rivet's own, and the Launch HN thread questioned both the premise and the AWS bill.
- 02GitHub Actions Lockfiles: What gh actions-lock Pins and What It Leaves LooseWorkflow lockfiles went generally available on October 28. The extension's own README lists four gaps, including Docker images and reusable workflows in other repositories.
- 03C++26 Reportedly Fails Its ISO Ballot, With Contracts the Sticking PointSecondary reports put the October 23 vote at 12 yes of 20 and nine no votes of 22 bodies. The numbers do not reconcile, and no ISO or WG21 record has been published to settle them.
- 04A Postgres Benchmark Found jsonb Read One Key 2.7 to 10.6 Times Faster, and Larger on Disk in 6 of 11 CasesChris van Eijk of Now-Next loaded 1 million small documents and 5,000 large ones into PostgreSQL 17.11 and 18.6. The speed result is consistent. The storage result depends on whitespace and the 2 kB compression threshold.