Follow Discord
Sweep 09 Oct 2026 · 17:27Z Build v2.1.296 517 read Stable v2.1.287 Latest v2.1.296 Next v2.1.296 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.296 ·

Forked skill and slash command results can pass through a formatter; async agent failures carry a cause

Skills can supply a formatForegroundForkResult hook to reshape foreground forked results, and failed background agents now record a cause

Group of 2 Under the hood No documentation found Internal Changes
JSON All of v2.1.296
Under the hoodTier: how much it should matter to you
4Useful: my rating, 1 to 5
3Signal: worth watching, 1 to 5
SkillsArea: what it touches
Internal ChangesKind: in v2.1.296,
What probably matters to youSection of the release

What

Skills and slash commands can run "forked", meaning in a separate side conversation whose result is handed back to the main one. When that runs in the foreground (you wait for it to finish), its result can now be reshaped before it is used:

  • formatForegroundForkResult is a new optional hook on skill definitions. A hook here is a piece of code the skill provides that Claude Code calls at a set moment. It receives the result of a foreground forked skill or slash command and can reformat it. Before, the result text was used as it was.
  • If that hook throws an error, the error is logged as "fork result formatter threw".
  • Agents running in the background (async agents) now carry a cause when they fail, such as stalled, error, or api_error:kind.

Why

For most people this is plumbing they will not notice. Skill authors gain a way to shape what a forked run hands back. A recorded failure cause makes it possible to tell a background agent that stalled from one that hit an error.

Read from
Names in the bundleformatForegroundForkResult

See this entry in the whole of v2.1.296 →

Feedback