Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@test/e2e/test-launchable-smoke.sh`:
- Line 223: SC2024 is triggered by redirecting output outside sudo for the
backgrounded invocation of brev-launchable-ci-cpu.sh; fix by running the
redirection inside the elevated shell so the file is created/owned under the
sudo context, e.g. replace the current command with a sudo -E bash -c "exec
\"$REPO/scripts/brev-launchable-ci-cpu.sh\" >\"$INSTALL_LOG\" 2>&1 &" form
(ensuring INSTALL_LOG is quoted), or if the original behavior of creating a
runner-owned log file is intentional, add a targeted ShellCheck suppression
comment (# shellcheck disable=SC2024) immediately above that line with a short
rationale referencing SC2024 and INSTALL_LOG.
- Around line 157-164: The script currently always calls pass "Pre-cleanup
complete (clone dir pre-seeded)" even if the preceding setup steps fail; make
the clone/pre-seed commands fail fast and only call pass on success: remove the
tolerant "|| true" on the git -C "$NEMOCLAW_CLONE_DIR" checkout -B main HEAD
command (or replace it with explicit error handling), ensure the git clone and
git -C ... remote set-url origin commands are checked for non-zero exit codes
(or enable a shell-wide "set -e" at the top of the script), and if any of those
commands fail invoke the script's existing failure handler (or exit non-zero)
instead of proceeding to call pass "Pre-cleanup complete (clone dir
pre-seeded)"; reference the NEMOCLAW_CLONE_DIR variable and the pass invocation
to locate the lines to change.

---

Nitpick comments:
In `@test/e2e/test-launchable-smoke.sh`:
- Around line 444-446: The variable TIMEOUT_CMD is assigned via the command -v
checks but never used, causing dead code and ShellCheck SC2034; remove the three
lines that declare/assign TIMEOUT_CMD (the variable and the two command -v
checks) from the script, or if a timeout wrapper was intended, replace the dead
assignments by actually using TIMEOUT_CMD where timeouts are invoked (e.g., use
"$TIMEOUT_CMD" before commands) or export a single explicit timeout command;
specifically locate the occurrences of the symbol TIMEOUT_CMD and delete the
assignment block if unused, or wire it into timeout invocations to eliminate the
dead-variable warning.

Verify each finding against the current code and only fix it if needed.

Inline comments:
In @.github/workflows/e2e-branch-validation.yaml:
- Around line 25-27: Update the header comment that currently references
"e2e-launchable-smoke" to the correct job identifier "launchable-smoke-e2e" so
the note matches this PR's context; search for the string "e2e-launchable-smoke"
in the header comment near the top of
.github/workflows/e2e-branch-validation.yaml and replace it with
"launchable-smoke-e2e" to avoid misrouting maintainers.

---

Nitpick comments:
In @.github/workflows/e2e-branch-validation.yaml:
- Around line 120-121: The concurrency group uses a per-run unique fallback
(github.run_id) so cancel-in-progress never dedupes non-PR workflow_call runs;
change the expression used in the concurrency group from
e2e-branch-validation-${{ inputs.pr_number || github.run_id }} to a stable
branch/ref fallback such as e2e-branch-validation-${{ inputs.pr_number ||
github.ref }} (or github.ref_name) so branch-based runs share the same group;
update the group string accordingly where "group: e2e-branch-validation-${{
inputs.pr_number || github.run_id }}" appears and keep "cancel-in-progress:
true".

Verify each finding against the current code and only fix it if needed.

Inline comments:
In @.github/workflows/nightly-e2e.yaml:
- Around line 1288-1294: The new job launchable-smoke-e2e must be added to the
dependency list for the report-to-pr job so the PR summary waits for it; update
the report-to-pr job's needs array (referencing the job name report-to-pr and
the new launchable-smoke-e2e job) to include "launchable-smoke-e2e" alongside
the existing needs entries so report-to-pr runs only after that job completes.
- Around line 1288-1334: Add a new path_instructions mapping in .coderabbit.yaml
associating the test file test/e2e/test-launchable-smoke.sh with the job name
launchable-smoke-e2e; locate the existing path_instructions block and add an
entry following the same pattern as other E2E jobs (map the file path to the job
identifier "launchable-smoke-e2e") so the notify-on-failure coverage includes
this new job.

Verify each finding against the current code and only fix it if needed.

Duplicate comments:
In `@test/e2e/test-launchable-smoke.sh`:
- Around line 157-164: The script currently tolerates failures during the clone
pre-seed (notably the checkout with `git -C "$NEMOCLAW_CLONE_DIR" checkout -B
main HEAD 2>/dev/null || true` and unchecked git commands) but still emits pass
via `pass "Pre-cleanup complete (clone dir pre-seeded)"`; change this to
fail-fast by removing the `|| true` from the checkout, checking the exit status
of the `git clone`, `git -C ... checkout -B main HEAD`, and `git -C ... remote
set-url origin` commands (or enable `set -e` at the top of the script), and if
any of these fail call your existing failure handler instead of `pass` so the
script does not report success when pre-seed steps failed.

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@test/e2e/test-launchable-smoke.sh`:
- Around line 149-150: The cleanup currently runs sudo rm -rf
"$NEMOCLAW_CLONE_DIR" with no validation; add a guard that validates
NEMOCLAW_CLONE_DIR before any rm: ensure it is set and non-empty, not "/" or "~"
or "$HOME", and (preferably) resides under an expected safe parent (e.g., a tmp
or workspace prefix) or matches a strict pattern; if the check fails, log an
error and skip deletion (exit non-zero or continue safely). Update both
occurrences that call sudo rm -rf "$NEMOCLAW_CLONE_DIR" (the instance shown and
the one at lines referenced 559-562) to perform this validation using the same
helper check/conditional before calling sudo rm -rf.
- Around line 1-3: The variable TIMEOUT_CMD is assigned but never used,
triggering ShellCheck SC2034; either delete the unused assignment or make it
actually used/exported. To fix, locate the TIMEOUT_CMD variable assignment and
either remove that line if no timeout behavior is needed, or export/use it where
timeouts are applied (e.g., replace direct calls to commands with "$TIMEOUT_CMD
<command>" or add export TIMEOUT_CMD so subshells can access it), ensuring the
symbol TIMEOUT_CMD is referenced after its definition.

---

Duplicate comments:
In `@test/e2e/test-launchable-smoke.sh`:
- Around line 157-164: The pre-seed git commands currently swallow failures (the
`git -C "$NEMOCLAW_CLONE_DIR" checkout -B main HEAD 2>/dev/null || true` line)
and still call `pass "Pre-cleanup complete (clone dir pre-seeded)"`, so update
the script to fail fast: remove the `|| true` from the `git checkout -B main
HEAD` call and ensure the subsequent `git -C "$NEMOCLAW_CLONE_DIR" remote
set-url origin "$NEMOCLAW_CLONE_DIR"` is checked for non-zero exit (or enable
`set -e` at the top), and only invoke the `pass` helper after both git commands
succeed; if either command fails, exit non-zero or call the existing failure
helper instead of printing pass.