fix(shell): detect the current shell from the process tree, not $SHELL

[max-sixty]

wt switch could tell a fish user "shell requires restart" because $SHELL (the login shell, say zsh) had the integration line while the shell actually in use didn't. current_shell() keyed everything on $SHELL, which names the login shell rather than the shell wt was invoked from.

Process-tree detection. shell::ancestor_shell() (src/shell/utils.rs) walks up from wt's parent to the nearest enclosing shell: Linux reads /proc/<pid>/stat; macOS parses one ps -Ao pid=,ppid=,comm= snapshot. macOS cannot use per-pid p_comm (libproc/sysctl): /bin/sh re-execs bash via /var/select/sh, so an sh script's p_comm reads "bash", while ps's comm preserves argv[0]. Verified live and pinned by test_probe_reports_invoked_name_for_sh. The walk passes through non-shell wrappers (git, sudo) and plain script interpreters (sh, dash); a known-but-unsupported interactive shell (tcsh, ksh) stops the walk so a shell further up can't claim the session. current_shell()/current_shell_name() consult the walk first and fall back to $SHELL / PSModulePath (Windows behavior unchanged; Git Bash sets $SHELL itself).

Hedged messaging. The reason string is now "shell integration installed but not active", and the hint presents restart as the likely fix with an escape hatch: "A shell restart usually activates shell integration; if it doesn't, ask an agent to debug with the docs @ https://worktrunk.dev/llms.txt" (the LLM docs index works for any agent; the /worktrunk skill exists only for plugin users). Post-install hints are unchanged (restart advice there is unconditional by construction). wt config show now prints Detected shell: <name> (process tree) next to $SHELL in the diagnostic gutter.

Also fixed along the way: shell_from_name requires a word boundary, so fishd/bashtop/numactl no longer classify as shells (numactl previously matched Nushell via the bare nu prefix); the post-install/uninstall restart hints compare Shell values instead of display strings, which never matched process-tree names like pwsh or zsh-5.9.

Testing. Integration tests spawn wt under a test harness whose real ancestry (nextest, cargo, the developer's shell) would leak in, so WORKTRUNK_TEST_PARENT_SHELL is pinned empty in STATIC_TEST_ENV_VARS (and in the two PTY helpers that mirror it by hand, which CI round 1 caught: locally the walk found the dev zsh, on CI the runner's bash) and set per-test to simulate ancestors. New integration tests cover the three detection outcomes (mismatch suppresses restart advice, configured ancestor shows the hedged hint, unsupported ancestor overrides $SHELL); unit tests cover the classifier, the walk itself (injectable process-table lookup: wrapper pass-through, transparency, cycle/depth/init terminations), and both OS probes against live processes; wt config show is snapshot-tested with an override ancestor and run once unpinned so the production walk executes end-to-end.

Known limits, accepted: the walk stops at the nearest supported shell even when it's a non-interactive script interpreter (bash -c wrapping wt reports bash), hence the hedged wording; the walk is bounded at 16 hops and any unreadable hop falls back to $SHELL.

This was written by Claude Code on behalf of max

🤖 Generated with Claude Code

[worktrunk-bot]

The detection logic reads carefully and the test matrix is thorough — mismatch, configured-ancestor, and unsupported-ancestor are all covered, and test (macos) passing confirms the ps-comm/argv[0] assumption the whole macOS branch rests on. The one thing holding this back is codecov/patch (red): two of the newly-added regions are never executed by the suite.

1. The real process-tree walk (detect_ancestor_shell, the #[cfg(unix)] loop). Every test sets WORKTRUNK_TEST_PARENT_SHELL, which returns before the walk, so parent_id() → process_name_and_ppid → the bounded loop never runs end-to-end. process_name_and_ppid and ancestor_from_name are each unit-tested in isolation, but the driver that stitches them together isn't. This is somewhat inherent to the override design; the cleanest coverage is probably a serialized unit test that remove_vars WORKTRUNK_TEST_PARENT_SHELL and calls detect_ancestor_shell() directly (asserting it returns without panicking / hanging — the result is environment-dependent), or accepting the gap as a known infra path.

2. The wt config show "Detected shell: … (process tree)" line (render_shell_status). Config-show tests pin WORKTRUNK_TEST_PARENT_SHELL="", so ancestor_shell() is None and the branch never renders. This one is cleanly and deterministically coverable — set the env var to a shell name and the line fires. Sketch modeled on test_config_show_powershell_detected_via_psmodulepath (adjust helper names/ordering to match; the override must come after configure_wt_cmd, which sets it to ""):

/// `config show` surfaces the process-tree shell as "Detected shell: <name>
/// (process tree)" alongside $SHELL.
#[rstest]
fn test_config_show_detected_shell_via_process_tree(mut repo: TestRepo, temp_home: TempDir) {
    repo.setup_mock_ci_tools_unauthenticated();
    let settings = setup_snapshot_settings_with_home(&repo, &temp_home);
    settings.bind(|| {
        let mut cmd = wt_command();
        repo.configure_wt_cmd(&mut cmd);
        repo.configure_mock_commands(&mut cmd);
        cmd.arg("config").arg("show").current_dir(repo.root_path());
        set_temp_home_env(&mut cmd, temp_home.path());
        set_xdg_config_path(&mut cmd, temp_home.path());
        cmd.env("SHELL", "/bin/bash"); // login shell differs from the ancestor
        cmd.env("WORKTRUNK_TEST_PARENT_SHELL", "zsh");
        assert_cmd_snapshot!(cmd);
    });
}

Happy to push a commit with the show.rs test (and a driver test if you want one) — just say the word. Leaving this unapproved only because codecov/patch is a merge gate here.

One minor, non-blocking observation on the boundary check is inline.