Follow Discord
Sweep 03 Nov 2026 · 22:27Z Build v2.1.318 536 read Stable v2.1.311 Latest v2.1.318 Next v2.1.318 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.318 ·

REPL mode respects plugin hooks, auto mode and an output size limit

In REPL mode, plugins that check tool calls can block REPL-made calls, queued scripts are judged by auto mode, and long output is cut to a size limit

You'll notice Improvements
JSON All of v2.1.318
You'll noticeTier: how much it should matter to you
4Useful: my rating, 1 to 5
4Signal: worth watching, 1 to 5
REPL ModeArea: what it touches
ImprovementsKind: in v2.1.318,
What probably matters to youSection of the release

Unclear It is not clear when Claude Code runs in REPL mode, so whether these changes reach you is unknown.

What

REPL mode is a way of working where Claude writes small scripts that call tools, instead of calling each tool one at a time. This release adds a large set of rules for it:

  • Certain hook events, including PreToolUse (the hook that runs before a tool call and can block it), now mark a plugin as one that may check tool calls
  • If such a plugin is installed, or Claude Code cannot tell whether one is, a tool call made from inside a script is refused with a message saying the call was not made
  • Queued scripts are checked by auto mode, the setting where Claude Code decides on its own which actions are safe to run
  • Output from a script that finishes later is cut once it passes a byte limit, with a note saying so
  • Shell commands run from a script get their results formatted, including text about saved output and background tasks

A separate server-controlled switch makes MCP tools (tools supplied by outside servers you connect) raise an error inside a script when they fail. Its built-in fallback, used only when the server sends nothing, is on.

Why

Tool calls made from inside a script can no longer slip past a plugin that is meant to inspect every tool call, and auto mode now applies to scripts too. Very long output from a script no longer floods the conversation.

Read from
Feature flag
tengu_repl_mcp_error_throw 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 v2.1.318: on

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 v2.1.318. It isn't a statement about your account. What a flag value here can and cannot tell you

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 doubtIt is not clear when Claude Code runs in REPL mode, so whether these changes reach you is unknown.
The name it cites is new in this buildNew in this build: tengu_repl_mcp_error_throw

See this entry in the whole of v2.1.318 →

Feedback