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 2026-11-19
stories 28

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

Prompt runs now ignore other people's comment edits

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

claude-code-action closed a documented gap between its security policy and its behavior, tightening which events can start a prompt run while the rest of the stack moved through routine version bumps.

Keep prompt runs to the actor's own comments and the step's output files in place anthropics/claude-code-action

by mhegazy

docs/security.md always claimed that editing or deleting another person's comment or review does not start a run, but that held only when track_progress was set. It now holds for every prompt: a prompt workflow will no longer fire on someone else's edited or deleted comment or review, while new comments and reviews trigger as before. This is a breaking change for any workflow that relied on the...

Keep prompt runs to the actor's own comments and the step's output files in place (#2098) anthropics/claude-code-action

by Mohamed Hegazy

The merged commit behind the same change. Note that label_trigger still does not apply to prompts run without track_progress, so label-gated prompt workflows behave exactly as they did.

Remove the unused prepare entrypoint (#2097) anthropics/claude-code-action

by Mohamed Hegazy

action.yml has routed to src/entrypoints/run.ts since a much earlier change, and nothing in the repository imported or ran src/entrypoints/prepare.ts, so it is gone along with the uncalled checkTriggerAction. The author states there is no user-facing change.

feat(discord): add list_emojis tool anthropics/claude-plugins-official

by eoger-ant

The Discord plugin gained a list_emojis tool taking chat_id and an optional case-insensitive query, returning entries as <:name:id> or <a:name:id> for animated ones, because react and reply need custom emoji IDs Claude previously had no way to look up. In a DM it reports only servers with an opted-in channel plus the application's own emoji, never every server the bot has joined.

Sync from claude-code-action base-action@f431ada anthropics/claude-code-base-action

by GitHub Actions

Routine maintenance elsewhere: claude-code-base-action synced from claude-code-action at base-action@f431ada, which bumped Claude Code to 2.1.328 and the Agent SDK to 0.3.328, the same CLI version claude-agent-sdk-python now bundles.

Quick answers

What shipped in Anthropic on November 19, 2026?
claude-code-action closed a documented gap between its security policy and its behavior, tightening which events can start a prompt run while the rest of the stack moved through routine version bumps. In total, 17 commits, 7 pull requests, and 4 releases landed.
Who contributed to Anthropic on November 19, 2026?
3 developers shipped this update, including GitHub Actions (publish-mods), eoger-ant, and Mohamed Hegazy.
What were the notable Anthropic updates?
Keep prompt runs to the actor's own comments and the step's output files in place, Keep prompt runs to the actor's own comments and the step's output files in place (#2098), and Remove the unused prepare entrypoint (#2097).