186 wires and counting

$ follow Anthropic

Keep up with Anthropic in about 3 minutes: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

fair warning: these emails are deeply technical. diffs, version numbers, CVEs, benchmark deltas. if that's not your idea of a good read, this isn't your newsletter.

Folds into your digest — weekly by default, monthly if you prefer. Unsubscribe in one click.

$ status

wire 2026-11-14
stories 51

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

A zstd decoding fix in Claude Code, and the SDK stops dropping your prompt's uuid

By RepoJournal · Filed · About Anthropic · Composed from the cited sources · methodology

Claude Code v2.1.325 patches a security flaw in zstd decompression and ships a policy-block error that tells you how to continue, while the Agent SDK finally surfaces the prompt uuid it was discarding.

v2.1.325 anthropics/claude-code

by claude-code-publisher[bot]

The release fixes "a security issue in how zstd-compressed responses are decoded," and error text for organization policy blocks now names the escape hatch: run /rewind to change the message, or /clear. Nothing in the notes says which versions are affected, so treat the upgrade as the fix.

feat(types): expose user_message_uuid and user_message_uuids anthropics/claude-agent-sdk-python

by qing-ant

A prompt sent while an injected turn is running can be answered inside that turn and come back carrying that turn's origin; the CLI sent the prompt's uuid on the result and the SDK threw it away. ResultMessage, AssistantMessage and StreamEvent now expose optional user_message_uuid and user_message_uuids, so apps can find their own answer instead of chasing the injected turn's.

Use the base branch's npm and Bun config on outside pull requests anthropics/claude-code-action

by mhegazy

Hooks, MCP servers and plugins on the base branch can start npm or Bun in the checkout, and those tools read config from the repository root, which until now came from the pull request. On pull requests from authors without write access, root config now comes from the base branch wherever it differs.

Close every pull request opened by a non-member anthropics/claude-plugins-official

by jordanecker-ant

The auto-close workflow kept letting outside pull requests through when they added repos already in marketplace.json; that exception and the advisory scope-guard check are both gone, and validate, scan and approval remain the required gates. knowledge-work-plugins got the same change.

rename(sap-fiori-mcp-server): sap-ux-fiori-tools, repoint to plugins-coding-agents/fiori anthropics/claude-plugins-official

by abibbs-ant

The long tail: SAP's rename lands as sap-ux-fiori-tools with the old name in the renames map, repointed from the stale packages/fiori-mcp-server pin (1.12.6) to plugin v1.16.16 on current main, and claude-plugins-official gains twenty animated theme-examples backdrops shipped as sample code rather than maintained themes.

Action items

Quick answers

What shipped in Anthropic on November 14, 2026?
Claude Code v2.1.325 patches a security flaw in zstd decompression and ships a policy-block error that tells you how to continue, while the Agent SDK finally surfaces the prompt uuid it was discarding. In total, 36 commits, 12 pull requests, and 3 releases landed.
Who contributed to Anthropic on November 14, 2026?
7 developers shipped this update, including claude-code-publisher[bot], GitHub Actions (publish-mods), qing-ant, dhollman, jordanecker-ant, abibbs-ant, and mhegazy.
What were the notable Anthropic updates?
v2.1.325, feat(types): expose user_message_uuid and user_message_uuids, and Use the base branch's npm and Bun config on outside pull requests.