Follow Discord
Sweep 22 Oct 2026 · 04:32Z Build v2.1.308 527 read Stable v2.1.308 Latest v2.1.308 Next v2.1.308 Feeds RSS JSON llms.txt llms-full.txt Unofficial
HomeReleasesv2.1.301entry #58

Claude Code v2.1.301 ·

A failed MCP tool call can stop Claude from ending its turn too early

If an MCP tool asks to end the turn while a tool call in the same batch failed, Claude Code can now keep the turn going

You'll notice Improvements
JSON All of v2.1.301
You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
3Signal: worth watching, 1 to 5
MCPArea: what it touches
ImprovementsKind: in v2.1.301,
ImprovementsSection of the release

Unclear Which kinds of MCP tool failure count, and what the added note says to Claude, are not stated.

What

MCP (Model Context Protocol) is the system Claude Code uses to connect to outside tools. Claude can run several tool calls together as one batch. An MCP tool can ask for Claude's turn to end.

Claude Code can now overrule that request and keep the turn going. It does this by adding a short note to the conversation for Claude. All of these must be true:

  • The request to end the turn came from an MCP tool.
  • No extra step budget has been used up.
  • This is not the last step Claude is allowed to take.
  • Another result in the same batch counts as a failure.

The behavior sits behind a switch that Anthropic controls from its servers.

Why

This changes whether Claude keeps working after an MCP tool call fails, instead of stopping while a failure is left unaddressed.

Read from
Feature flag
tengu_cozy_fern Not enough to say

Nothing here resolved what this flag was doing on this version, so nothing here should be read as on or off.

This account: no value returned · anonymous baseline: no value returned · compiled default in v2.1.301: off

These values were read against a different version of Claude Code, so treat them as the nearest reading available instead of one taken on this release.

Read once, for one account on one subscription tier, against v2.1.301. It isn't a statement about your account. What a flag value here can and cannot tell you

What has happened since
Flag reading moved The flag server now returns off for tengu_cozy_fern, read as this account. A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.
Flag reading moved The flag server now returns nothing at all, having stopped carrying it for tengu_cozy_fern, read as this account. A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.
Flag reading moved The flag server now returns off for tengu_cozy_fern, read as this account. A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.
Flag reading moved The flag server now returns nothing at all, having stopped carrying it for tengu_cozy_fern, read as this account. A reading is one sample. Claude Code evaluates its flags remotely, so no client sees the targeting rule behind a value and this says nothing about your account.

See this across every release →

How sure we are
Something disagreesSomething we can check disagrees with this entry, or the writer said they could not settle it.
The writer flagged doubtWhich kinds of MCP tool failure count, and what the added note says to Claude, are not stated.
The name it cites is new in this buildNew in this build: tengu_cozy_fern

See this entry in the whole of v2.1.301 →

Feedback