Claude Code Attribution Setting: Turn It Off (v2.1.281)

Looking for the Claude Code attribution setting that finally turns off those Co-Authored-By lines? The short answer is just below — it’s one line in settings.json, it’s official, and it shipped in v2.1.281 on 23 September 2026. Here’s exactly how it works, where to put it, and when I’d actually leave attribution on.

What the Claude Code attribution setting does

By default, Claude Code signs its work: commits get a Co-Authored-By trailer and pull request descriptions get a “Generated with Claude Code” footer. Plenty of teams are fine with that. Plenty aren’t — client repos, open-source projects with strict commit conventions, or simply a lead dev who wants clean git history.

Until now the workarounds were ugly: post-commit hooks that rewrote messages, CLAUDE.md instructions the model had to remember to follow, or hand-editing before pushing. As of v2.1.281 (23 September 2026), the official changelog adds a first-class switch: "attribution": false in settings.json, which the changelog describes as hiding commit/PR attribution.

How to use the Claude Code attribution setting

Three steps:

1. Update Claude Code to v2.1.281 or later (claude --version to check).
2. Pick your scope. Project-wide and shared with the team: .claude/settings.json in the repo. Just you, everywhere: ~/.claude/settings.json.
3. Add the setting:

{
  "attribution": false
}

That’s the whole job. New commits and PRs created by Claude Code stop carrying the attribution lines. If you manage settings across a team, this slots into the same file as your permissions and hooks — same place you’d configure things like the output style.

Should you actually turn attribution off?

My honest take: it depends on who reads your git history.

• Turn it off on client deliverables where commit conventions are contractual, on repos where trailers break tooling, or on personal projects where it’s just noise to you.
• Leave it on in team repos where reviewers triage AI-assisted commits differently, and in open-source where disclosure builds trust. The trailer is genuinely useful metadata when you’re debugging “who wrote this and why” six months later.

The point of v2.1.281 is that it’s now a deliberate choice in config rather than an accident of defaults — which is exactly how team-level AI policy should work.

The rest of v2.1.281 (and why it’s an enterprise-flavoured release)

The attribution switch travelled with a clearly enterprise-leaning changelog. Also in v2.1.281, per the official changelog:

• Desktop policy blocks: blockReadsOutsideWorkingDirectories and disableBypassPermissionsMode for managed Claude Desktop deployments.
• Cross-account AWS: assume_role on Claude apps gateway Bedrock upstreams — the gateway calls Bedrock as an IAM role assumed through STS, in another AWS account if needed, optionally one session per developer.
• Bedrock guardrails: a guardrail: {id, version} option on Bedrock upstreams.
• Plugin hygiene: MCP server checks added to claude plugin validate — handy if you’ve been building plugins since the plugin eval command landed.
• Auto mode: a recommendation surfaced in /insights.

Worth noting for context: the day before, v2.1.280 (22 September 2026) made Claude Opus 5.5 the default model, and earlier in the week v2.1.275 added the send-now key. Anthropic is shipping daily at the moment — the changelog is worth a skim every Monday. See my post on Opus 5.5 becoming the Claude Code default for what that switch means in practice.

The bottom line on the Claude Code attribution setting

If you’ve been stripping Co-Authored-By lines by hand or with hooks, delete the hack: update to v2.1.281 or later and set "attribution": false in settings.json. It’s official, it covers commits and PRs, and it scopes cleanly per-project or per-user. Every claim above comes from the official Claude Code changelog entry for v2.1.281, dated 23 September 2026 and fetched while writing this guide.

FAQ: Claude Code attribution setting

How do I stop Claude Code adding Co-Authored-By to commits?

Add “attribution”: false to your settings.json. The setting shipped in Claude Code v2.1.281 on 23 September 2026, per the official changelog, and hides commit and PR attribution.

What version of Claude Code has the attribution setting?

Claude Code v2.1.281, released 23 September 2026. Check your version with claude –version and update if you’re older than that.

Does the attribution setting affect pull requests too?

Yes — the official changelog entry says the setting hides commit and PR attribution, so both the commit trailer and the generated-with footer in PR descriptions are covered.

Where do I put the attribution setting?

In settings.json — project-level (.claude/settings.json, shared with your team via git) or user-level (~/.claude/settings.json, applies to everything you run locally).

Is it OK to remove Claude attribution from commits?

It’s your repository and now an officially supported setting, so mechanically yes. Whether you should depends on your team’s disclosure norms — many teams keep attribution on shared repos precisely so reviewers know which commits were AI-assisted.

What else shipped in Claude Code v2.1.281?

Per the official changelog: desktop policy blocks (blockReadsOutsideWorkingDirectories and disableBypassPermissionsMode), assume_role for cross-account AWS Bedrock access via STS, Amazon Bedrock guardrails support, MCP server checks in claude plugin validate, and an auto mode recommendation in /insights.

Related reading

• Claude Code plugin eval command
• Claude Code send-now key
• Claude Opus 5.5 is now the Claude Code default

Last updated September 2026. This is the living guide to Claude Code attribution setting — it gets updated as the tools change.