Lane 10 PR 7a: the Test the bot pages (#736) #902

Merged

claude-bot merged 3 commits from build/lane10-7a-test-the-bot-pages into main 2026-09-28 03:24:59 +00:00

Conversation 1 Commits 3 Files changed 12 +1351 -7

claude-bot commented 2026-09-28 01:30:51 +00:00

Collaborator

Copy link

Part of #736 (lane 10). Frontend only, on main. It's built on PR 6's existing /recording-tests routes, with no new endpoints.

What it builds

• Settings → Recording, "Test the bot" (#testbot, inside "Starting a recording", as drawn):
  • you choose to keep or delete the results;
  • "no Discord" is known before you press, and every other refusal gives the server's own sentence;
  • the running stages poll every 2 s: recording, handing over, transcribing, summarising;
  • the result shows each stage's tick and timing, what it heard, and the summary.
  • A failure's title follows the drawing: "Stopped at transcription", "Stopped at the summary", "Nothing heard".
  • It also lists the kept test runs, with a confirmed delete.
• The session page: while Start is refused as too early (D149), a "Test the bot ›" link opens that section (D222).
• D223:
  • an administrator of this server reads "you run this server", with a link to Admin → Discord for a bot refusal or Admin → AI for a transcriber failure;
  • everyone else reads the named-administrators sentence.
  • A summariser failure is the campaign's own setting, so it reads the same for everyone.

Departures

1. Only "no Discord" is known before the press. "Bot not set up" and "bot too old" arrive after it, in the server's words, and nothing is recorded by a refused press. The campaign payload carries no bot-reachability flag.
2. The transcript is one paragraph, not per-speaker lines, because the server returns a single string.

Checks

• vitest: 200 files, 2670 passed. eslint: 0 errors.
• Tests: unit tests for the state machine and wording, including both readers of D223, plus the page and RecordingPanel tests.
• Render check: I looked at the shots myself. Two faults were fixed: the failure titles, and the result chip wrapping at narrow widths.
  • ⚠️ The phone width isn't confirmed yet. Headless Edge on Windows crops rather than narrows below about 500 px. This PR waits for a real-390 pass on a running stack before it merges, together with #900.

🤖 Generated with Claude Code

Part of #736 (lane 10). Frontend only, on main. It's built on PR 6's existing `/recording-tests` routes, with no new endpoints. ## What it builds - **Settings → Recording, "Test the bot"** (`#testbot`, inside "Starting a recording", as drawn): - you choose to keep or delete the results; - "no Discord" is known before you press, and every other refusal gives the server's own sentence; - the running stages poll every 2 s: recording, handing over, transcribing, summarising; - the result shows each stage's tick and timing, what it heard, and the summary. - A failure's title follows the drawing: "Stopped at transcription", "Stopped at the summary", "Nothing heard". - It also lists the kept test runs, with a confirmed delete. - **The session page:** while Start is refused as too early (D149), a "Test the bot ›" link opens that section (D222). - **D223:** - an administrator of this server reads "you run this server", with a link to **Admin → Discord** for a bot refusal or **Admin → AI** for a transcriber failure; - everyone else reads the named-administrators sentence. - A summariser failure is the campaign's own setting, so it reads the same for everyone. ## Departures 1. **Only "no Discord" is known before the press.** "Bot not set up" and "bot too old" arrive after it, in the server's words, and nothing is recorded by a refused press. The campaign payload carries no bot-reachability flag. 2. **The transcript is one paragraph,** not per-speaker lines, because the server returns a single string. ## Checks - **vitest:** 200 files, 2670 passed. **eslint:** 0 errors. - **Tests:** unit tests for the state machine and wording, including both readers of D223, plus the page and `RecordingPanel` tests. - **Render check:** I looked at the shots myself. Two faults were fixed: the failure titles, and the result chip wrapping at narrow widths. - **⚠️ The phone width isn't confirmed yet.** Headless Edge on Windows crops rather than narrows below about 500 px. **This PR waits for a real-390 pass on a running stack** before it merges, together with #900. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude-bot added 2 commits 2026-09-28 01:30:52 +00:00

feat(webapp): Test the bot, in Settings and on the session page (#736, lane 10 PR 7a) ... 3a4ebde770

Builds D124's "Test the bot": the launch (30s, stoppable early, keep-or-
delete), the running stages, the result (per-stage timing, transcript,
summary, a plain failure sentence naming who can fix it) and the kept
test runs, under Settings -> Recording, retention & AI, polling the
existing /recording-tests routes. The session page's recording panel
offers a link to it while Start is refused for being early (D149, D222).

New: src/api/recordingTests.js, src/pages/recording-test/*.
RecordingPanel now takes `routed` as a prop (matching RecordingSection's
existing pattern) rather than calling useNavigate() itself, so tests that
render it standalone don't need a Router.

Known departures from the drawing (see TestTheBot.jsx's header comment):
"the bot isn't set up" / "the bot needs updating" surface only after a
press, since the campaign payload carries no such flag yet; D223's
administrator-aware wording is not implemented (same sentence for every
reader).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

feat(webapp): D223, an administrator's own wording for Test the bot's server-matter failures ... 62253707e0

An administrator reads "you run this server" with a link into the right
part of Admin; everyone else keeps the server's own sentence, which
already names the administrators. Applies to:
- a start refusal that is the server's to fix (not set up, unreachable,
  or too old -> Admin -> Discord), never "nobody's in the voice channel"
  or "the bot is busy", which are about the moment;
- a failed test's transcriber stage (Admin -> AI).

The summariser failure is left as it is for every reader: it already
names the campaign's own settings, which is the GM's to fix whether or
not they also administer the server (flagged in the handback for the
coordinator to overrule if they want it treated as a server matter too).

testTheBot.js: isServerMatter/BOT_TOO_OLD_TEXT (mirrors the backend's
own refusal shape by status/text, since the frontend has no other way
to tell "needs updating" apart from "busy"), failureAdminLink,
failureSentence's new isAdmin parameter.

TestTheBot.jsx now reads useAuth() for is_admin, which needed
useAuth.jsx mocked in CampaignSettingsRecording.test.jsx (TestTheBot
mounts inside that page) and TestTheBot.test.jsx, following the
existing CampaignSettingsArchive.test.jsx pattern.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

rbrooks referenced this pull request from a commit 2026-09-28 02:02:51 +00:00

fix(webapp): Test the bot's card and Test runs get room of their own (#902)

rbrooks added 1 commit 2026-09-28 02:02:51 +00:00

fix(webapp): Test the bot's card and Test runs get room of their own (#902) ... 2a40914157

On a phone the running card sat flush against the window field's hint above
it and "Test runs" butted onto its bottom edge; neither neighbour brings a
margin. Seen in the real-390 render check.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude-bot commented 2026-09-28 02:03:15 +00:00

Author

Collaborator

Copy link

The phone check is done, on a real seeded stack. Playwright ran inside the stack's container at a true 390×844 and at 1440, in light and dark, covering: launch, a real running test against the fake bot, a real finished result, and the session page's "Test the bot ›" link. Every shot passes scrollWidth <= innerWidth, and I looked at them myself.

One fault fixed (2a409141): on a phone, the running card sat flush against the field hint above it, and "Test runs" butted onto the card's bottom edge. The fix adds vertical margins, scoped to this component.

**The phone check is done, on a real seeded stack.** Playwright ran inside the stack's container at a true 390×844 and at 1440, in light and dark, covering: launch, a real running test against the fake bot, a real finished result, and the session page's "Test the bot ›" link. Every shot passes `scrollWidth <= innerWidth`, and I looked at them myself. **One fault fixed (2a409141):** on a phone, the running card sat flush against the field hint above it, and "Test runs" butted onto the card's bottom edge. The fix adds vertical margins, scoped to this component.

claude-bot merged commit 63c20a9b0e into main 2026-09-28 03:24:59 +00:00

claude-bot referenced this pull request from a commit 2026-09-28 03:25:01 +00:00

Merge PR #902: lane 10 PR 7a, the Test the bot pages (#736)

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

backend

Issue related to the backend

  

bot

Issue related to the bot

  

bug

Something is not working

  

duplicate

This issue or pull request already exists

  

enhancement

New feature

  

frontend

Issue related to the frontend

  

help wanted

Need some help

  

invalid

Something is wrong

  

question

More information is needed

  

recording

Issue related to the recording system

  

security

Security hardening / vulnerability work

  

wontfix

This won't be fixed

No labels

backend

bot

bug

duplicate

enhancement

frontend

help wanted

invalid

question

recording

security

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

rbrooks/Quest-Board!902

No description provided.