You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Part of #823.
Spec: docs/superpowers/specs/2026-10-08-event-logs-workbench-ui-handoff-design.md (see the section references below).
Design: Claude Design canvas "Event Log Viewer Redesign" (board exports to be attached).
Spec: §14.2
Scope (2 files)
e2e/fixtures/event-log-data.ts (new): TS port of the mockup generator (seed 1337, 25 templates,
~1,331 events; Security as an accessDenied coverage gap); synthetic analysis-session and diagnosis payloads.
Placeholder user names only.
e2e/screenshots/capture.spec.ts: event-logs captures (table, selected + dock, grouped, timeline,
charts, correlation) using window.__e2e_ipc_overrides__ for the evtx_* commands listed in §14.2.
Acceptance
npm run screenshots -- -g event-logs produces deterministic PNGs at 1440×900 @2x.
Fixture contains no real machine names, SIDs, tenant IDs, or user names.
Before Phase 3 the captures show the current UI (baseline for comparison).
PR #880 is merge-ready at d6de74c5. CI is green at that head (16 of 16 check runs, including the always-on engine drift test), CodeRabbit approved that exact commit after a full review, and a clean final charter review is posted on the PR (#880 (comment)). The review viewed both committed PNGs and sampled the fixtures: every identifier is synthetic. #880 and #873 (also merge-ready) rewrite the same doc comment in playwright.screenshots.config.ts, and #873 adds the host-path guard (#892). Merge them in either order; Main then adds a follow-up commit to the second one that resolves the comment and calls assertNoHostPaths before the two Event Logs captures, and takes it through CodeRabbit and a final review again.
PR #880 is merge-ready again at 3a64d692. It took main after #873, adds the host-path guard before both Event Logs captures, and recaptured the PNGs with the 24px status bar (checked: synthetic identifiers only). CI is green at that head (16 of 16), CodeRabbit approved that exact commit after a full review, and a clean final charter review is posted on the PR (#880 (comment)).
Activity
adamgell commentedon Oct 9, 2026
Lane started: branch
feat/event-logs-screenshot-harness, cut from main at 1d9a2b0.The captures go in a new file,
e2e/screenshots/event-logs.spec.ts, instead ofcapture.spec.ts.capture.spec.ts. Both changes append to the end of the file, so they would conflict.testDiralready includes the new file, sonpm run screenshots -- -g event-logspicks it up.The fixture ports the mockup's
wbBuildgenerator: seed 1337, 25 templates, timestamps from a fixed base date.user01, per spec §3.4.accessDeniedcoverage gap.Per the acceptance criteria, the captures show the current UI as the baseline before Phase 3.
test(event-log): env-gated capture of real analysis-session replies (#…
test(event-log): replay real engine replies in the screenshot harness (…
chore(screenshots): Event Logs baselines from real engine replies (#828)
test(event-log): reuse the record helper and name paths in capture-te…
test(event-log): pin the screenshot replay to its exact engine input (#…
test(event-log): recapture engine replies with input and source hashes (
test(event-log): replay committed input in an always-on drift test (#828
test(event-log): record clean provenance for the replies (#828)
fix(e2e): settle dev server ownership before event log captures (#828)
fix(e2e): prove dev-server identity with a nonce, pin the input hash (#…
test(event-log): record clean provenance for the replies (#828)
docs(e2e): drift test docs name every ignored and pinned field (#828)
test(event-log): record clean provenance for the replies (#828)
fix(e2e): publish the engine input only after the capture succeeds (#828
fix(e2e): publish engine fixtures with same-directory renames (#828)
docs(e2e): capture script header matches the staged publish (#828)
adamgell commentedon Oct 10, 2026
PR #880 is merge-ready at
d6de74c5. CI is green at that head (16 of 16 check runs, including the always-on engine drift test), CodeRabbit approved that exact commit after a full review, and a clean final charter review is posted on the PR (#880 (comment)). The review viewed both committed PNGs and sampled the fixtures: every identifier is synthetic. #880 and #873 (also merge-ready) rewrite the same doc comment inplaywright.screenshots.config.ts, and #873 adds the host-path guard (#892). Merge them in either order; Main then adds a follow-up commit to the second one that resolves the comment and callsassertNoHostPathsbefore the two Event Logs captures, and takes it through CodeRabbit and a final review again.test(e2e): host-path guard before the Event Logs captures; recapture …
adamgell commentedon Oct 10, 2026
PR #880 is merge-ready again at
3a64d692. It took main after #873, adds the host-path guard before both Event Logs captures, and recaptured the PNGs with the 24px status bar (checked: synthetic identifiers only). CI is green at that head (16 of 16), CodeRabbit approved that exact commit after a full review, and a clean final charter review is posted on the PR (#880 (comment)).test(event-log): synthetic Event Logs screenshot harness (Phase 2, #828…