Skip to content
Agents tracked: 258 Downloads (7d): 219M up 6.1% GitHub stars: 5.5M VS Code installs: 148M Releases (7d): 293 Agent status: 2 with issues Updated Oct 7, 2026

VoltAgent changelog: what's new each month

Every stable VoltAgent release summarised by month: the highlights, new features, improvements, fixes and anything you need to act on. 3 months covered; the current month updates daily.

September 2026

1 release: 2.11.0

September brought a focused update to VoltAgent with improved flexibility for custom workspace sandboxes.

Highlights

  • Custom workspace sandboxes can now receive original executecommand commands without normalization.

Improved

  • Toolkit command normalization can be disabled for custom workspace sandboxes, allowing access to raw command arguments.

Fixed

  • Custom workspace sandboxes now properly receive unmodified executecommand commands and arguments.

August 2026

2 releases: 2.9.2 → 2.10.0

August brought updates to VoltAgent's Model Context Protocol support and template handling. Two releases focused on improving type safety and preserving template values.

Highlights

  • Upgraded Model Context Protocol SDK to 1.30.0 with stateless Streamable HTTP handling
  • Added type-safe MCP annotations and metadata support for tools
  • Fixed preservation of JavaScript replacement tokens in template variables

New

  • Stateless Streamable HTTP handling with startHTTP option
  • Serverless and request-scoped streaming options for HTTP
  • Type-safe MCP annotations for tools
  • MCP metadata forwarding through the server adapter

Improved

  • Model Context Protocol SDK updated to version 1.30.0
  • Tool annotations now include metadata and type information

Fixed

  • JavaScript replacement tokens are now preserved in simple template variable values

July 2026

2 releases: 2.8.1 → 2.9.0

July brought improvements to usage tracking consistency and fixed a critical workflow state bug that was corrupting input data.

Highlights

  • Usage metrics now clearly distinguish between final-step and aggregate usage across all steps in agent outputs.
  • Fixed workflow state corruption where finish() was overwriting stored input with final output.
  • Usage reporting now consistent across all AI provider implementations.

New

  • Exposed AI SDK usage semantics in agent onEnd outputs with separate final-step and aggregate tracking.

Improved

  • Aggregate finish usage now handled consistently across all supported providers.

Fixed

  • Workflow state.input was being overwritten with final output when finish() was called.

Summaries are written automatically from the official release notes (full changelog ↗); check the original notes before relying on a detail. VoltAgent: pricing, features and alternatives · All changelogs