Follow Discord
Sweep 22 Oct 2026 · 21:04Z Build v2.1.309 528 read Stable v2.1.308 Latest v2.1.309 Next v2.1.309 Feeds RSS JSON llms.txt llms-full.txt Unofficial
HomeReleasesv2.1.302entry #106

Claude Code v2.1.302 ·

MCP sign-in and rejected-credential messages adapt to where Claude Code runs

MCP authentication messages now come from shared helpers and can say to start a new task or chat instead of pointing to /mcp

Group of 4 You'll notice Improvements
JSON All of v2.1.302
You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
MCPArea: what it touches
ImprovementsKind: in v2.1.302,
ImprovementsSection of the release

Unclear Which sessions get the new-task wording instead of /mcp is not stated.

What

MCP servers are outside tools that Claude Code connects to. When one needs you to sign in, or turns down the credentials it was given, Claude Code shows a message. These messages changed:

  • The messages for a rejected headersHelper credential (a command you configure to produce login headers) and for a rejected Authorization header in a server's config now come from shared helpers, helperRejected and headerRejected, instead of being written in place. The old wording told you to check the helper, or update the header, and then run /mcp to reconnect. The new wording was not visible in the findings.
  • When a particular check is true, MCP messages leave out /mcp and tell you to start a new task or chat instead.
  • The reminder Claude receives when MCP servers need sign-in now chooses its wording by whether those servers are claude.ai connectors. In some cases it tells Claude to explain the step "without naming a command or settings page".

Why

Where /mcp cannot be run, advice to run it is a dead end, and these messages now give a step you can take instead. The exact wording you see may differ from earlier versions.

Read from
Names in the bundle/mcp
Since it was published

The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.

Added since A small documentation edit on How the agent loop works touched a line naming /mcp after this was published. | **Tool definitions** | Every request for core built-in tools such as Bash, Read, and Edit; on demand for MCP tools and other built-in tools, which [tool search](/docs/en/agent-sdk/mcp#mcp-tool-search) defers by default | Every schema loa… agent-sdk/agent-loop see the edit
Confirmed since Anthropic's documentation has since written up /mcp, on Automate actions with hooks. * MCP tools marked [`requiresUserInteraction`](/docs/en/mcp#require-approval-for-a-specific-tool) hooks-guide see the edit
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 sessions get the new-task wording instead of `/mcp` is not stated.
Anthropic's documentation agreesAnthropic's documentation has since written up /mcp, on Automate actions with hooks.
Anthropic's release notes agreeFixed Claude in the desktop app pointing you to claude.ai settings or /mcp to authorize or reconnect an MCP server where that option isn't…

See this entry in the whole of v2.1.302 →

Feedback