[lawrencecchen]

Summary

Base red (priority zero). On feat-cmux-next, cmux-tui cli::app::tests::share_chat_waits_for_the_owners_answer (src/cli/app/tests.rs:359) fails with "expected one action.run call". Base job b973957eb8a382f63c3d57c0.

Culprit: ff01dcc (#30550), the commit that added the test. No later commit changed the share chat path: git log ff01dcc58da..origin/feat-cmux-next shows no commit to cli.rs, cli/app.rs, cli/app/ or cli/share_cli.rs. The test asked app::parse for share chat --surface 1, but share is not and never was in APP_SCOPES, so parse answers Ok(None). The test was red from the commit that added it (fleet run at ff01dcc below).

The product is right, so this fixes the test, not the product. cmux share chat is not a mux command, so cli.rs run_app_action_fallback runs it. share_cli::action leaves it as typed (it rewrites only share revoke|kick), and app::run_cli_action sends one action.run for share chat. That run waits PERSON_RUN_TIMEOUT, because run_action takes waiting_run_timeout("share chat"). The test now takes that same path. It checks that parse answers None, that share_cli::action answers None, and that run_action("share chat", ["--surface", "1"], Cli) is one action.run with args.surface = "1" and the person deadline. The waiting_run_timeout assertions stay as they were.

FYI @teamleaderleo: this is your #30550 test.

Testing

cmux-ci run --class isolated --os linux --script scripts/ci/cmux-tui-rust-check.sh --ref SHA --arg test --arg share_chat_waits_for_the_owners_answer:

• Red at the base 13fd3f48bcb0: FAIL share_chat_waits_for_the_owners_answer, "expected one action.run call". Step 66f5715c3c0e7dd8111fc935.
• Red at the culprit ff01dcc58da (the commit that added the test): the same FAIL and the same message. Step 010e0c9a4036dd5e5fe2845b (Blacksmith run manaflow-ai/cmuxterm-hq 41421315070). So no later commit broke it.
• Green at the head a6dbfc45352, 5 of 5: steps 4f5a3a7d9b98c40cd2ff24be, 5ee2d80352be204faf7a7a35, a871d0e5b3593cf69b304a5d, 71275c72030cee25ef3ec2e9, c62b5b8c0a7e879719525869 (each 1 passed).
• Panic ratchet (--arg panic) at the head: pass, step 7fd6769aa0f14554aeed7d25.

Non-green checks at the head are base findings, not caused by this PR. clippy (changed cmux-tui packages): too_many_arguments on cmux-tui-core snapshot_written, from #31931; fixed by #32092 (merged). defaults-snapshot (cmux-next guards) is a base finding. Its failing step is "Menu rows run registered actions" on AgentPaneSessionInfoContent.swift, from #31930. The same check is red on unrelated open PRs, and #32072 and #32073 fix it.

Observed / Not run

• Observed: the fleet step logs above (nextest PASS/FAIL lines); python3 scripts/cmux-next/prepush-gate.py passes at the head.
• Not run: no local build or test on this VM (fleet only). The CLI was not run end to end against a live app (cmux share chat with an app socket); the claim that the product sends one action.run comes from reading cli.rs and cli/app.rs, not from a live run.

Changelog

none

[github-actions]

cmux status _{(one comment, edited in place; each line is updated by the job that owns it)}

Review a6dbfc4: HQ feed card · gallery before/after runs · checks

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=pr32065 -f ref=base-red-share-chat -f platform=macos -f backend_mode=local

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.4142821761901 (937b52c218)
• cmux NEXT nightly 1.0.0-nightly.4143019763401 (17e607bf23)
• cmux NEXT nightly 1.0.0-nightly.4143038272801 (912566a91a)
• cmux NEXT nightly 1.0.0-nightly.4143282790301 (50298238a3)

[lawrencecchen]

Review (subagent, correctness first): SHIP at a6dbfc4

Read-only review. Nothing compiled or run on this machine.

What I checked:

• Product path for cmux share chat --surface 1 (cli.rs run): none of the run_if_requested handlers or remote_host::handles (needs host first) takes share. run_app_scope gets Ok(None) from app::parse because share is not in APP_SCOPES. The mux grammar (parse_command, shorthand, command::parse) does not know share, so it errors and run_app_action_fallback runs. There, agent_view::action returns None (needs agent view). share_cli::action returns None for any verb except revoke and kick. --surface is not a global flag, so it stays in command_args. The fallback then joins the words before the first flag (share chat). new_window_spelling does not match. run_cli_action calls run_action(name, rest, ActionName::Cli) one time, which builds one action.run with wait: true, cli: true and args: {"surface": "1"}. Its timeout is waiting_run_timeout("share chat"), which is PERSON_RUN_TIMEOUT (210 s). call adds one idempotency key and sends that one request. Busy retries happen only when the app says the run never started, and they use the same key. Result: exactly one action.run "share chat" with the person timeout. The product is correct. Only the old test was wrong.
• The new test follows the same steps: parse returns Ok(None), share_cli::action returns None, then run_action("share chat", &words[2..], ActionName::Cli). It also checks method, action, args and timeout. It fails if share chat leaves PERSON_ACTIONS.
• Compile, by reading: use super::* brings in parse, run_action, ActionName, AppCommand, PERSON_RUN_TIMEOUT, WAITING_RUN_TIMEOUT, waiting_run_timeout and json! (bare json! is already used at tests.rs:47). super::super::share_cli::action is pub(super) in a private child module of cli, so cli::app::tests can see it. Its return type Option<Result<(String, Vec<String>), String>> has PartialEq and Debug for assert_eq!(.., None). &words[2..] is &[String], which matches run_action. Value has Display for "{params}".
• History: git log -S' "share",' -- cli/app.rs finds no commit, so share was never in APP_SCOPES. At ff01dcc, APP_SCOPES is the same list as on head. git log ff01dcc58da..HEAD on cli.rs, cli/app.rs, cli/app/run.rs, cli/share_cli.rs, cli/app/tests.rs and cli/app/wait.rs shows only this PR's commit. The claim is true: the test was red when cmux NEXT: a guest link an agent or script asks for waits for the owner's Allow (#30263) #30550 added it and nothing broke it later.

Non-blocking gaps:

• The test copies the fallback's steps by hand. It does not prove that the mux grammar still rejects share (if a mux share scope is added, the fallback stops running) or that the word split before the first flag gives share chat. run_app_action_fallback needs a socket, so a unit test cannot call it directly. This is acceptable for a base-red repair.

[lawrencecchen]

Non-green checks at a6dbfc4, classified:

• cmux-next daemon tests (same-tree) (run 41422401667): 2 red Swift tests in ClassicUpgradeSimulationTests.swift:239,263 (anOffSSHHostDoesNotHoldBackLocalWorkspaces, theOrderIsReadOnlyOnceQueuedPlacementsLand). This PR changes one Rust unit test in cmux-tui/src/cli/app/tests.rs and no Swift or daemon code, so these reds are not caused by it.
• clippy (changed cmux-tui packages): base red from fix(cmux-tui): a host reconnect ends the output store's tail, so a kill right after it restores the gap's output #31931 (too_many_arguments on snapshot_written), fixed by fix(cmux-tui-core): snapshot_written takes the snapshot's end as one value (base red: clippy too_many_arguments) #32092 (merged).
• defaults-snapshot: base red from cmux NEXT: draw Session Info with native AppKit glass #31930 (inline menu row in AgentPaneSessionInfoContent.swift), fixed in fix: route native Session Info menu rows through ActionRegistry #32072 / Session Info: row menu runs through the shared menu-action helper, not an inline selector (base red) #32073.
• next-autoland / verdict: wrapper of the reds above.

[lawrencecchen]

This PR left a cmux-tui clippy red on feat-cmux-next: unused_qualifications at cmux-tui/crates/cmux-tui/src/cli/app/tests.rs:352 (std::path::Path::new). Fixed forward in #32128.