[JacobStephens2]

Summary

Command record + Session traces
  → shared span reader
      ├─ show <command> --spans / --span <id> --json
      └─ Waterfall tree → summary Peek
           Command → Runs → stages → Sessions → calls / sub-agents

Build the settled #995 layout: lazy Session expansion, ranked slowest calls, shared timeline and zoom, keyboard navigation, and a summary Peek at laptop and 390 px widths. Preserve live browsing state and existing Command details. Recorded Session totals take precedence over their children's repeated usage.

Closes #996

Evidence

• Before: show widgets#7 --spans --json rejected the new flags. The Spec review's empty, chips, and totals reproductions each failed on the reported behavior.
  After: all three retained reproductions pass, including the reviewer's independent reruns; the targeted browser run passed all 9 tests (tree/Peek/zoom, refresh, legacy timing, Stop, and the three regressions).
• cargo test --test console: 58 passed, 1 fixture entry ignored. cargo test --test command_surface --test session_traces: 28 passed, 2 fixture entries ignored.
• cargo clippy --all-targets -- -D warnings: passed.
• Real retained Spec: Session traces, OTLP JSON Lines in place of Session logs #700 Command: 2,342 spans, including 2,227 tools; unique IDs and no missing parents. Laptop and 390 px screenshots inspected locally.
• Final full Console browser suite: 71 passed, 1 skipped, including the last-row phone Peek regression.
• CI passed on final head 92139a5: full cargo test, formatting, strict clippy, Python tooling, Console and site browser checks. Release artifact checks also passed.
• scripts/affected-tests issue-700-branch-2 selected full Rust validation. Its duplicate local nextest run was intentionally stopped after exact-head CI passed, following the repository's CI-reuse rule: 1,170.99 s elapsed including compilation (20.75 s). The interrupted local run is not counted as a passing suite.

Testing seams

• Public CLI tree/detail JSON and real Session recording through the binary.
• Console HTTP routes paired with that CLI data.
• Served browser interactions at laptop and 390 px widths, including live changes and retained regressions.

Review coverage

Standards: 2 design heuristics addressed. Spec: 3 behavioral findings reproduced and fixed, with their tests retained. Reports are saved locally in .thirdshift-review-ROPib1/standards.md and .thirdshift-review-ROPib1/spec.md, each including the initial review, one follow-up, and the final files-read inventory.

Changed file left unread by both independent review axes: src/console/assets/look.js. Its one-line activity selector extension followed their coverage pass; the retained Theme motion browser regression covers it. No generated-output exclusions.

Unaddressed findings

Standards

None. Both findings were accepted and addressed.

Spec

None. All three findings were accepted and addressed; no finding was deferred to the Day shift or a Spec decision.

Merge Danger

Door: two-way

Reversible reader and Console changes, with additive metadata on the first live Session event.

Blast Radius: Console

Span rendering, trace inspection, and Theme activity indicators.

Built with codex · gpt-6.1-sol · xhigh