[JacobStephens2]

Summary

Report elapsed wall time for a top-level Run or Spec run on stderr before its final outcome, using the existing Run notification duration format.

command starts → sample monotonic clock
command ends
  skipped Pass / pure review → existing output
  child Run    → existing outcome
  top level    → Took: 1h 2m 3s → existing outcome/PR line

Closes #51

Evidence

• Before: cargo test --test progress_lines ends_a_successful_run_by_naming_the_pr failed: expected one elapsed summary, found zero.
  After: The same test passes, with one elapsed line immediately before the PR outcome and only the PR URL on stdout, with notifications disabled.

• Failure and graceful interruption checks preserve exit code 1 and the original cause.

• A linear Spec run produces one elapsed summary; its two child Runs produce none.

• Deterministic duration cases cover fractional seconds, minute/hour boundaries, and durations beyond 24 hours.

• Final validation: cargo test (1783 passed, 2 ignored), cargo fmt --check, and cargo clippy --all-targets -- -D warnings passed.

Test seams

• CLI command stdout, stderr and exit status, using real Git repositories and fake external Harness/GitHub executables: Run/Spec run endings, child Run suppression, failure, interruption, disabled notifications and skipped Pass output.
• Shared progress duration formatter: injected Duration values at second/minute/hour boundaries, preserving the Run notification's existing format.

Unaddressed findings

Standards: None.

Spec: None.

Security: None.

Security review outcome

No unaddressed introduced findings.

Review coverage

• Security: Guidance-mode review of git diff b58d24d287aa956679f59341482eaaf709dc9a64...HEAD and supporting command lifecycle, logging, notification and child-output code found no introduced security findings. Read the relevant attack-class guidance; no repository SECURITY.md or dedicated threat model was present. Offline checks in a network-isolated sandbox passed: 43 unit tests, all 11 progress_lines integration tests, and the linear Spec run test.
• Standards: S1 (duplicated elapsed assertions) addressed with shared test support. No remaining findings; unread changed files: none.
• Spec: SPEC-1 (elapsed output on pure Architecture/Security passes) reproduced with the reviewer's exact command, fixed, and retained as a regression test. The exact command failed before and passed after. No remaining findings; unread changed files: none.

Reports are retained in .thirdshift-review-wZmfRL/standards.md and .thirdshift-review-wZmfRL/spec.md in the Run's worktree.

Merge Danger

Door: two-way

Blast Radius: terminal

Adds a stderr summary to top-level Run/Spec endings. Existing outcomes, stdout URLs, exit codes and notification content are preserved.

Built with codex · gpt-6.1-sol · high