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
HomeReleasesv2.1.299entry #159

Claude Code v2.1.299 ·

Clearer messages when --tools or --disallowed-tools cannot block MCP tools

--tools now explains it cannot remove MCP tools, and cloud sessions warn when a --disallowed-tools connector wildcard blocks nothing

Group of 2 You'll notice Improvements
JSON All of v2.1.299
You'll noticeTier: how much it should matter to you
2Useful: my rating, 1 to 5
1Signal: worth watching, 1 to 5
Cloud SessionsArea: what it touches
ImprovementsKind: in v2.1.299,
ImprovementsSection of the release

What

--tools and --disallowed-tools are flags (options given when starting Claude Code) that limit which tools Claude may use. MCP tools come from MCP servers, outside programs that add extra tools, and claude.ai connectors are MCP tools linked through your claude.ai account. Two messages are new:

  • --tools and MCP tools: an entry in --tools that starts with the MCP prefix is now rejected with an explanation: --tools limits built-in tools only, and an MCP tool stays in the session unless --disallowed-tools names it. Before, such an entry got a generic rule or name-format message. Names starting with the claude.ai connector prefix are now accepted by --tools validation.
  • Connector wildcards in cloud sessions: when you launch a cloud session, any --disallowed-tools rule with a * in a claude.ai connector's name is reported as not applied. The note says such a rule blocks no connector in a cloud session and suggests one rule per connector, such as mcp__claude_ai_Slack. Before, these rules were not checked at all. Claude Code also now counts connector-name rules by kind (both names, local name only, wildcard) for its usage statistics.

Why

Both cases used to fail quietly: you could believe a tool was blocked when it was not. Now you are told, and if you rely on wildcard rules to block connectors in cloud sessions, you should switch to one rule per connector.

Read from
Names in the bundle--disallowed-toolsmcp__claude_ai_Slack
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.

Confirmed since Anthropic's documentation has since written up --disallowed-tools, on Error reference. * Fix the rule at its source, which the full warning names in parentheses: a settings file path, or the flag itself for `--allowed-tools` and `--disallowed-tools`. A `claude-settings-<hash>.json` path that doesn't exist on disk stands for … errors see the edit
How sure we are
One source agreesOne thing we can check says the same as this entry.
Anthropic's documentation agreesAnthropic's documentation has since written up --disallowed-tools, on Error reference.

See this entry in the whole of v2.1.299 →

Feedback