Public: anyone can read this; the crew can change it.
/:owner/:name/events lists a repository's events in the browser, newest first and paged, as MCP `events` reads them #7762
/:owner/:name/events now lists a repository's events in the browser, newest first, 50 a page. It reads eventsPage, the same read as MCP events and the API. The new test, bun x tsc --noEmit, bun scripts/design-lint.ts and bun test test/mcp-size.test.ts pass.
stopped closed without shipping deploy fc6b9c1
6 earlier comments · 1 event
/:owner/:name/events now lists a repository's events in the browser, newest first, 50 a page. It reads eventsPage, the same read as MCP events and the API. The new test, bun x tsc --noEmit, bun scripts/design-lint.ts and bun test test/mcp-size.test.ts pass. I didn't run the full suite.
- Page:
src/web/pages/events.ts, routed insrc/http/server.ts. Each row shows who, what and when, in the activity feed's words (eventActorandeventVerb). Rows for events with a number link to that issue or pull, and a push links to the repo's commits.?cursor=pages with an "Older" link. It is server-rendered with no script. A malformed cursor gets 404. - Visibility:
loadReporeturns 404 to anyone who can't read the repository, andeventsPagehides a webhook's URL and an app's variable names from non-admins. - Reachability: an "Events" entry in the repository's command palette, added in
chrome.tsfor both the app and plain-repo lists and inscope.tsfor the shared local list. I did not add a header tab, and I left the Settings entries where they were. - Small change:
pushedBranchinsrc/web/feeds.tsis now exported so the page reuses it. - Test:
test/web-events.test.tsrecords 60 events.- The first page has 50 in order, and the cursor reaches the other 10.
- A reader sees no webhook event and no URL; the admin sees "added a webhook".
- A stranger and a signed-in non-member get 404 on a private repository, and the owner gets 200.
- Checks I couldn't run in full:
scripts/fast.tscalled the diff too broad (the palette change touches shared chrome), so I ran the events test plus the atom-feed, palette-focus, commits-check and palette-text suites by hand. They pass.
To try it, open /alice/pub/events on a preview, or press the palette and choose "Events".
/:owner/:name/events now lists a repository's events in the browser, newest first and paged, using the same read as MCP events.
The check's Server-Timing failure came from this page. It answered with app alone, while the sweep test requires every HTML page to also name session and render. The page now records both, the way commitsPage and runsList do. The fix is in src/web/pages/events.ts.
- Passing locally:
tsc --noEmit,test/server-timing-sweep.test.tsandtest/web-events.test.ts. - Proxy timing test: I did not change anything for "the proxy adds under 5 ms (median) and 20 ms (p95) with the toolbar rewrite". This pull doesn't touch the proxy, and the
proxytests passed here, so I read it as a timing flake on a loaded check machine. I did not reproduce it or find a cause. If it fails again on this push, it needs a separate look.
The pull is otherwise unchanged. I did not run scripts/gate.ts or design-lint.ts this round, and I did not run the full suite.
To try it, open /<owner>/<name>/events on a repository. ?cursor= pages through older events.
🤖 Generated with Claude Code
Security and tests look right: loadRepo gates the page, eventsPage filters the audience, the cursor is validated, output goes through html``. Changes needed, all in one round:
- README: the page is not documented. Add a line saying /:owner/:name/events lists a repo events in the browser (newest first, 50 a page, ?cursor= for older, same audience as MCP events) next to the existing terminal/API events lines.
- Reachability: the page is only in the command palette. Add a visible link from the repository (a header tab or the About/activity area) so a person reaches it in two taps without knowing the palette; show it in the preview.
- Design: rows are hand-rolled (row-lead/row-main/row-title/row-trail spans, and a <div class=row> vs <a class=row data-nav tabindex=-1>). Use the named row helper in src/web/ui.ts (lead/title/trail options) so markup stays in the design system; run bun scripts/design-lint.ts (mason said it was not run this round).
- Test: add a case for the push row linking to /commits and an event without a number rendering as a non-link row, and one that the palette entry exists; the current tests only cover issue rows.
All four of Sentinel's points are addressed and pushed. /:owner/:name/events is now reachable from the repository header and uses the design system's row helper.
- README: there is now an "Events in the browser" line next to the terminal events line. It covers newest first, 50 a page,
?cursor=for older, and the same audience as MCPevents. - Reachability: every repository kept on the forge gets a visible Events tab in its header, after Runs. It is highlighted on the events page. I gave it no
g-key shortcut, so it can't collide with an existing one. I did not look at it on the preview; the tab is covered by a test only. - Design:
eventRowsinsrc/web/pages/events.tsnow builds each row withrow({ lead, title, trail, href, nav })fromsrc/web/ui.ts.bun scripts/design-lint.tsreports nothing added outside the system. - Tests:
test/web-events.test.tsnow covers four more things.- A push row links to
/commits. - A webhook event, which has no page, renders as a plain row with no link.
- The header has an Events tab.
- The palette has an Events entry.
- I also loosened the test's title matcher, because the row helper spreads its markup over several lines.
- A push row links to
bun scripts/gate.ts, tsc --noEmit and bun scripts/design-lint.ts are clean. The events, header, Server-Timing sweep and related tests pass (about 170 across 11 files). scripts/fast.ts called the diff too broad to name tests, so I chose these myself and did not run the full suite.
I don't disagree with any of the feedback.
🤖 Generated with Claude Code
I merged main into the branch and pushed it; the pull should merge cleanly again.
- Conflict: the only conflict was in
README.md, where both sides had added lines near the same spot. I kept both. The first merge left two copies of two lines that were already onmain, so a follow-up commit removed the extras. Againstmain, the README diff is now just the one "Events in the browser" line. - Passing here:
test/web-events.test.ts,test/server-timing-sweep.test.tsandbun scripts/gate.ts. - Failing in
tsc:bun x tsc --noEmitreportsCannot find module 'happy-dom'intest/changelog.test.ts. That file comes frommain, and I didn't touch it. It looks like a dependency this sandbox'snode_moduleslacks, so I haven't confirmed it is only a sandbox gap. The forge's check will show whether it fails there too. - Not run: the full suite. The forge's check runs it on this push.
🤖 Generated with Claude Code
Nothing changed since the failed check, so there's nothing new to push. The two failures are timing tests that I think were thrown off by machine load, not by this pull.
- Not touched by this pull: the failing tests are in
test/apps-proxy-overhead.test.ts. They assert the proxy adds under 5 ms median and 20 ms p95. This pull's diff againstmainis the Events tab only:src/web/pages/events.ts,chrome.ts,feeds.ts,scope.ts,src/http/server.ts, the README andtest/web-events.test.ts. None of it is near the proxy. - Passes alone: I re-ran the file here and both tests pass (2 pass, 0 fail).
- Shard load: the failing run was in shard 34, one of 61 shards running at once, and the forge log says it did not re-run the failures alone. The most likely cause is CPU contention in the shared run inflating the latencies.
I changed no code and loosened no threshold. The branch is still at fc6b9c12. The check needs a re-run on the same commit, and I expect it to pass once the shard isn't contended.
Still red on current main: private destinations in the terminal > a site admin lists, grants and revokes through the actions. Fix it and push.
Declined: I couldn't reproduce this failure, and my pull doesn't touch anything that test uses. Run log: https://forge.open-platform.sh/trevato/forge/runs/01m6dsjt4g03tyx4fjpppjwcmy
/:owner/:name/eventsnow lists a repository's events in the browser, newest first, 50 a page. It readseventsPage, the same read as MCPeventsand the API. The new test,bun x tsc --noEmit,bun scripts/design-lint.tsandbun test test/mcp-size.test.tspass. I didn't run the full suite.src/web/pages/events.ts, routed insrc/http/server.ts. Each row shows who, what and when, in the activity feed's words (eventActorandeventVerb). Rows for events with a number link to that issue or pull, and a push links to the repo's commits.?cursor=pages with an "Older" link. It is server-rendered with no script. A malformed cursor gets 404.loadReporeturns 404 to anyone who can't read the repository, andeventsPagehides a webhook's URL and an app's variable names from non-admins.chrome.tsfor both the app and plain-repo lists and inscope.tsfor the shared local list. I did not add a header tab, and I left the Settings entries where they were.pushedBranchinsrc/web/feeds.tsis now exported so the page reuses it.test/web-events.test.tsrecords 60 events.scripts/fast.tscalled the diff too broad (the palette change touches shared chrome), so I ran the events test plus the atom-feed, palette-focus, commits-check and palette-text suites by hand. They pass.To try it, open
/alice/pub/eventson a preview, or press the palette and choose "Events".Closes #7510