[teamleaderleo]

Workspaces now record whether a person or auto-naming chose their name, the same way tab names already do (NameSource { User, Auto } and name_revision). The daemon is the only writer, and it decides whether an automatic rename may land. This is the daemon half that the merged app lane (#21441, #21445) already calls; until now the app fell back because no daemon served workspace-name-source-v1.

• Storage. RegistryWorkspace and the mux Workspace gain name_source and name_revision, stored in two additive workspaces columns. Rows from before the migration read as user at revision 0, so names people already set stay pinned. Both commit paths re-check the stored row's name, source and revision inside the transaction, so a late auto-namer callback loses.

• Rename rules. rename-workspace and workspace.rename accept source, expected_generation, expected_name_revision and release. An automatic rename is refused if:

  • a person set the current name (operation.failed, reason_code: name_pinned);
  • the new name is empty (name_empty);
  • the name revision moved since the caller read it (revision.conflict).

  A rename without source is still a plain user rename, as before. source: "auto" with release: true hands a pinned name back to auto.

• Projection and capability. The raw tree, list-workspaces and the resource extra show name_source and name_revision, and the daemon advertises workspace-name-source-v1.

• CLI. cmux workspace <ws> rename --source auto --expected-generation G --expected-name-revision N, and --release-to-auto.

• App capability list. DaemonCapabilities moves workspace-name-source-v1 from unservedByBundledDaemon to optional, so the capability can't ship unserved again. plans/cmux-next/daemon-capabilities.json and the DaemonCapabilityExportTests ceiling are updated to match.

The code comes from #20710, carved onto current feat-cmux-next. Only #20710's own hunks are included in files it shares with #21322.

Freeze WINDOW areas: CORE (cmux-tui-core, the protocol spec and regenerated SDK bindings, the cmux-tui CLI). It also touches the CmuxNext app's daemon capability list (Packages/macOS/CmuxNext/.../DaemonEndpoint.swift, plans/cmux-next/daemon-capabilities.json) and one Native/OptChat/optchat-chief struct literal. Not LOCK: no Cargo.toml or Cargo.lock changes. The merge goes through the Air Blue coordinator, who can read the WINDOW.

Why 105 files

• 43 files are regenerated SDK bindings (cmux-tui/bindings/*) and their per-language resource descriptor hashes; one more is the Rust conformance adapter, which builds the new request fields.
• About 25 are two-line name_source / name_revision additions at each Workspace / RegistryWorkspace constructor in mux/ and the registry tests.
• The rest is the feature itself, its tests, the spec (commands.md, resource-operations-v2.json, sdk-schema.json), cmux-app.d.ts, one optchat-chief struct literal that gains the new request fields, and the Swift capability list change.

Testing

Local Linux run (Zig 0.16.0, ghostty-next submodule, user-level libclang, a sparse checkout widened to the files the crates read):

• cargo test -p cmux-tui-core --no-fail-fast on cd97e404a19: 2474 passed, 1 failed, 10 ignored across 17 targets. The one failure, server::terminal_snapshot::socket_tests::a_burst_of_grid_changes_collapses_into_the_last_grid, is timing-sensitive and passes alone.

  • After the god-file move (f0f48d1c447): --lib gives 2388 passed and 1 failed. That failure is a different timing test, browser::tests::reconfigure_and_attach::loaderless_navigation_retries_snapshot_invalidated_by_same_document_event, and it passes alone.

• cargo test -p cmux-tui --no-fail-fast on cd97e404a19: 2878 passed, 8 failed, 6 ignored across 75 targets. None of the 8 come from this change:

  • 5 also fail alone on the base a7428da446a (base-red): every_safe_transport_operation_has_a_noun_first_path (session.journal.append live_reply), daemon_started_shells_run_the_bundled_cli_after_user_startup_files, workspace_list_in_the_app_answers_the_sidebar_order, the_documented_customize_template_installs_as_is and split_close_and_restore_survive_a_daemon_restart.
  • 2 pass when run alone; they failed under load (Too many open files, timing): provider_reconnect_replacement_commits_as_provider_active and remote_mux_binding_rejects_a_stale_requested_owner_socket.
  • 1 reads design/agent-icons/manifest.json, which is outside this sparse checkout: every_hook_provider_is_in_the_agent_brand_manifest.

• What the first run found on f016fceef1e, fixed in cd97e404a19: 8 failures in tests that landed on feat-cmux-next after feat(next): workspace name source and revision for auto-naming (hq#1957) #20710's base. 6 were real:

  • The journal decorates stored workspaces with extra.name_source/name_revision, but projected workspace values had no extra. The public fold never matched, so every commit restated every workspace (4 batch_close tests and projected_effect_holds_writer_fence_through_commit_and_publishes_once). Every workspace value builder now carries the same extra.
  • Creating a workspace at a key bound to another live workspace hit the name fence before the identity check (key_close_cannot_rebind_a_live_durable_workspace_identity). The fence now applies only to the same workspace.

  The other 2 were sparse-checkout misses.

• This slice's own new tests (resource_name_tests.rs, workspace_registry/workspace_name_tests.rs, resource_router/topology/workspace_mutations/workspace_name_tests.rs, server/tests/workspace_name_source.rs, cli/command/tests/name_source_tests.rs) all pass in these runs.

• Formatting: cargo fmt --check for cmux-tui-core and cmux-tui, and rustfmt --check on the conformance adapter and optchat-chief/src/workspaces.rs: clean.

• Bindings: bindings/codegen/generate.py --check for cpp, java, python, rust, typescript and zig: clean. Go: there's no Go toolchain on this host, so the byte-exact gofmt check was skipped. I generated Go with an identity formatter and compared the token stream (whitespace, ; and trailing commas normalized) to the committed files; all five match, and the manifest hashes come from the committed bytes. CI's generator check is the first byte-exact gofmt check.

• Spec and app checks: check-resource-api-boundary.py, check-sdk-schema.py, check-spec-inventory.py and their 78 unittest cases; check-app-capability-references.py, check-daemon-capabilities-fresh.py and tests/daemon-capabilities-fresh.test.sh; check-no-godfiles.sh --only rust --base a7428da446a on f0f48d1c447. All pass.

• Not run: Swift tests, and check-daemon-capabilities.sh, which needs a bundled daemon. CI runs both.

Changelog

none

🤖 Generated with Claude Code

[github-actions]

Receipt for f0f48d1 (this PR's head; a new push replaces it)

• Visual review: HQ feed card · gallery before/after runs
• Checks: this head
• App of this head (app.zip, kept 1 day): gh workflow run reload-build.yml -R manaflow-ai/cmux --ref feat-cmux-next -f tag=pr21648 -f ref=split/20710-workspace-name-storage -f platform=macos -f backend_mode=local

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: Repository: manaflow-ai/cmux/.coderabbit.yaml
• Review profile: ASSERTIVE
• Plan: Advanced
• Run ID: f8ddac9a-1502-422c-b85a-a2d52da2cd05

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Not verified: macOS jobs did not run: compile and app tests were skipped on f0f48d1c44; ci-status is green without them.

CI passes on f0f48d1c44 (run 39137983690 attempt 1).
macOS jobs did not run: compile and app tests were skipped: Fast static checks skipped (the macos job needs both to succeed). Re-run CI once they pass, before merging.

Written by scripts/ci/classify_failures.py (ci-failure-attribution.yml); signatures are its SIGNATURES table. A machine verdict is the runner's fault, not this PR's; yours means the failing file is one this PR changes, also red on main that main's latest full suite fails the same way, seen on other PRs that it failed on another pull request's run lately.

[teamleaderleo]

Bead cx-l8fth created for this PR by the cmux coordinator.

[cursor]

Bugbot needs on-demand usage enabled

Bugbot uses usage-based billing for this team and requires on-demand usage to be enabled.

A team admin can enable on-demand usage in the Cursor dashboard.