Release Notes

Dr. Olivia goes back to your textbooks mid-conversation New

Until now, Dr. Olivia pulled the relevant pages from your course materials once — based on the first question you asked. That works until the conversation moves, which it always does: you ask a follow-up two levels deeper, or you change subjects entirely, and she's still holding the pages from where you started.

She can now go back and search your textbooks again, mid-answer, as many times as a conversation needs — and she writes the search herself, in the proper terminology, rather than being stuck with however you happened to phrase the question. You don't have to ask her to; it just happens.

Three things this changes in practice:

• Follow-ups go deeper. "Walk me through the actual mechanism" now sends her back for the mechanism, instead of re-explaining the overview she already had on hand.
• Changing subjects works. Switching from renal to cardiology halfway through a session no longer leaves her teaching from the wrong chapter.
• She checks herself against your material. Before stating something your materials don't clearly support, she goes and reads what they actually say.

This search is deliberately limited to your textbooks — not lecture captions or video summaries. Those are someone's interpretation of the material; the textbook is the material. When the point is to confirm a fact, she checks the source.

If the answer genuinely isn't in your curriculum, she'll tell you that plainly rather than filling the gap from general knowledge and leaving you to wonder which is which.

Video sessions start again on computers with a webcam Fixed

Starting a video session could fail immediately — Dr. Olivia never appeared and the session simply didn't connect. It affected computers that have a working camera, which is most of them; oddly, machines with no camera at all were unaffected.

The cause was in how we set up the video connection. When we meant "use the normal camera settings," we were passing along an empty value instead of leaving the setting alone, and the video service rejected the whole request before the session could open. We now leave the setting untouched, which is what was intended. Nothing changed about how the camera behaves once you're in a session, and machines without a camera keep working exactly as before.

Now open to our partner schools New

Dr. Olivia is no longer just for PHSU — students and faculty at our partner institutions can now sign in with their own school's Canvas. On the sign-in page, choose your school from the “Select your school” picker (search by name or code) and continue with Canvas as usual. If you arrive from a link inside your school's Canvas — or from your school's own Dr. Olivia address — your school is detected automatically and you can skip the picker entirely. Each school authenticates against its own Canvas, so you only ever see your institution's courses and materials.

See every server's settings at a glance Admin New

The System admin page has a new Configurations section that lists every server in the pool, not just the one that happened to answer your request. Pick a server from the dropdown to see what that node actually resolved at startup — its SMTP settings, build, and whether each critical key is set. When the settings aren't identical across nodes, a warning banner points you straight at the odd one out, so a server that missed a configuration change is now obvious instead of a guessing game. Secrets are never shown — passwords and keys appear only as “set” or “not set”.

Reorganized System page Admin Improved

The System page now has a left sidebar — Host, Caches & Vectors, Configurations, and Tooling — so you can jump straight to a section. The content area scrolls on its own while the sidebar and header stay put.

Smoother external-user management Admin Improved

Creating or editing an external user now shows a spinner while it saves and a confirmation toast when it's done — no more wondering whether your click registered. The page header also shows the external-user login link with Copy and Open buttons, so you can hand it to a new user directly.

Prompt updates roll out without a manual refresh Admin Improved

When Dr. Olivia's teaching prompts are updated, every server now notices the change on its own and reloads within about half a minute — no more clicking Refresh Prompts on each node, and no more wondering whether one server is quietly running an older copy. The System admin page's prompt-cache view now shows the loaded version, the version on disk, and a drift flag, so you can confirm at a glance that every node is current. A change to the avatar's spoken style still takes effect on the next new conversation, since each conversation sets up its voice when it starts.

Security: patched a bundled component Internal

Updated the bundled MessagePack serialization library to the patched 2.5.301 release, clearing a known high-severity advisory (CVE-2026-48109). Nothing changes in how the app works — a routine dependency security bump.

Quiz is now a student-only feature Faculty

When you're signed in as faculty, the Quiz button is now hidden on both the avatar and the text chatbot, and Dr. Olivia no longer launches the interactive student quiz overlay. Ask her to “generate a question” or “write a practice question” and she'll write it out inline in the conversation — ready to review or copy — instead of starting the timed student quiz experience.

End your chat session and leave feedback New

The text chatbot now has an End Session button in the top-right of the header, just like the avatar. Selecting it takes you to a quick rating screen for the session before returning you to the avatar selection page. Clicking the PHSU logo now does the same — on both the chatbot and the avatar it routes you through the feedback screen on your way back to selection. You can always skip the rating if you'd rather head straight back.

Textbook figures now show their source New

When Dr. Olivia shows you a real figure from one of your course textbooks, a small credit line now appears beneath it — the book title, author, edition, figure number, chapter, and page (whatever is known for that figure). The same credit appears under the figure in your post-session summary email, and where the book has a source link the credit is clickable. AI-generated images are unaffected: only real textbook figures carry a credit.

Open a generated image in a new tab New

When you enlarge an image Dr. Olivia generated for you, an Open in new tab button now lets you pop it out into its own browser tab to save or view full-size. (Textbook figures show their publisher citation instead, since those come from copyrighted source material.)

Ask for a different picture and get a different picture New

If an image isn't what you wanted, just tell Dr. Olivia — "that's not right", "show me a different one", "add labels", or "now the abnormal version". She no longer hands back the same picture: she finds a different matching textbook figure if one fits, and otherwise draws a fresh one for exactly what you asked. Refine as many times as you like.

"Draw me" vs. "show me" — your words pick the source New

Ask to be shown something ("show me…", "what does … look like", "describe…") and Dr. Olivia looks for the real textbook figure first. Ask her to create one ("draw me…", "sketch…", "illustrate…", or anything very specific and custom) and she draws it fresh instead of pulling a stock figure. Phrase it the way you want it.

Fixed: images no longer flash a blank box while loading

A figure's thumbnail now shows a loading placeholder until the picture has fully downloaded, instead of briefly collapsing into an empty space.

Ask for a picture — real textbook figures come first New

When you ask Dr. Olivia to show, draw, or visualize something, she now checks your course textbooks first. If a real, instructor-vetted figure already illustrates what you asked for, you'll get that actual image instead of an AI-generated one — so it matches your source material exactly. If no textbook figure fits, she falls back to reusing a previously generated image, and only creates a brand-new one when nothing suitable already exists. You don't have to do anything differently: just ask for the image, and she picks the most accurate source automatically.

Figures from your textbooks New

When a diagram from your course textbooks helps explain a concept, Dr. Olivia can now show you the actual figure — right alongside the explanation in the chat, and again in your post-session summary email. These are the real, instructor-vetted images from the source material, so what you see matches your books. As before, Dr. Olivia shows at most one figure per reply, and only when it genuinely reinforces the point.

A roomier way to switch tutors Improved

The tutor picker no longer takes up space on the page. Your current tutor now appears in the top-right corner — tap it to open a quick pop-up and choose a different one. Your choice is still remembered for next time, and Dr. Olivia remains the default.

Summary emails name the right tutor Improved

Your post-session summary email now labels each reply with the tutor who actually gave it, so a session led by another tutor reads correctly instead of always showing “Dr. Olivia.”

Rate your session New

When you end a session, Dr. Olivia now shows a quick rating screen: give the session a star rating (1–5) and, optionally, a short comment. Prefer to skip it? One click takes you back to the home page. The same rating link is also included at the bottom of your post-session summary email, so you can leave feedback later without signing back in. Your ratings help us keep improving the tutoring experience.

External sign-in for non-Canvas guests New

Dr. Olivia can now be accessed by people who don't have a Canvas account — such as prospective partners and invited guests — through a dedicated email-and-password sign-in. Administrators create these accounts, choose whether each guest gets the Student or Faculty experience, and can lock, reset, or remove them at any time. Guests can reset their own password and update their email (with verification) from a self-service account page. Passwords are stored only as salted one-way hashes. Existing Canvas and LTI sign-in is unchanged.

Choose your tutor New

When more than one tutor is available, the Selection page now shows an avatar picker so you can choose who guides your session. Dr. Olivia remains the default, and picking is optional — your choice is remembered for next time until you change it. Each tutor has their own voice and introduces themselves by name; everything else about the session works exactly the same.

Canvas LTI 1.3 launch for partner institutions New

Dr. Olivia can now be installed in a partner institution's Canvas as an LTI 1.3 (LTI Advantage) tool — students and faculty launch her right from inside a Canvas course. The launch keeps the normal Canvas sign-in, so Dr. Olivia still reads the student's own course materials on their behalf, and everything downstream (avatar, chat, materials, quizzes) works exactly as before.

Each partner is configured independently, and faculty are identified per institution. Existing PHSU sign-in is unchanged. Canvas admins: see the LTI 1.3 Setup guide.

ChEMBL drug lookups for pharmacology New

Dr. Olivia now reaches into EMBL-EBI's ChEMBL database for drug pharmacology — the layer that sits next to PubChem and PubMed. Ask how a drug works, what it targets, what it's approved for, or what clinical phase it's in, and she'll pull the live record — mechanism of action, molecular target (resolved to a readable name like "Cyclooxygenase"), indications, and development phase — and drop it in your Materials panel.

Three references now divide the work by the shape of your question, and she picks automatically: ChEMBL for what a drug does, PubChem for what a molecule is (structure, formula, SMILES), and PubMed for what the research says. Ask something that spans two and you get both, side by side.

ChEMBL is fully open — no API key or setup required. And because the record carries the molecule's SMILES, the same "now draw it" trick that works after a PubChem lookup works here too. Full walkthrough in the ChEMBL Lookups guide.

Rate Dr. Olivia's responses New

Every response from Dr. Olivia now has thumbs-up and thumbs-down buttons next to the timestamp. Click thumbs-up when a response was genuinely helpful, thumbs-down when it missed the mark, or click the active thumb again to clear your rating. Works in both the text chatbot and the avatar's transcript panel.

Ratings are stored per response, so we can spot patterns in what's landing well versus what isn't — whether that's a topic she consistently nails, a question type she struggles with, or a particular Socratic-vs-Faculty tone preference. There's no public leaderboard or visible score; the ratings feed our internal quality review only.

PubChem lookups for biochem New

Dr. Olivia can now reach into NCBI's PubChem database the same way she already reaches into PubMed — but for compounds instead of papers. Ask for a molecular formula, weight, SMILES, IUPAC name, or any physicochemical property (LogP, polar surface area, hydrogen-bond donors/acceptors) and she'll pull the live record and drop it in your Materials panel.

The two databases divide the work naturally: "what does the research say about hydrogen bonding" still routes to PubMed (it's a paper question), while "what's the molecular formula of hydrogen peroxide" routes to PubChem (it's a compound question). You don't have to name the database — she picks based on the shape of the question.

Best part: when you then ask her to draw the structure she just pulled, she feeds PubChem's SMILES into the image generator. That produces a dramatically more accurate chemical structure than the un-grounded version, especially for molecules bigger than three atoms.

Full walkthrough in the PubChem Lookups guide. Your existing PubMed personal API key (if you set one up) is reused for PubChem — no second key needed.

Dr. Olivia waits longer before responding Improved

Several students reported Dr. Olivia jumping in too early — especially when they paused mid-question to look back at the material or gather their thoughts. The avatar's pause detection has been tuned to be less eager: short thinking pauses (looking down at a question, re-reading a stem, deciding how to phrase a follow-up) are no longer treated as "I'm done talking." She'll wait until you've actually finished before she starts speaking.

Tradeoff to know about: when you are finished, there may be a slightly longer beat before she replies. If that ends up feeling sluggish in real use, the sensitivity is now a config knob we can nudge back up per replica without another deploy.

Generating images no longer get stuck loading forever Fixed

Some students reported the animated image placeholder spinning indefinitely — the diagram never appeared and the gray scanning box never went away. This happened when the connection briefly dropped during the 1–2 minutes an image takes to render: the finished image's "I'm ready" signal was sent while the page was momentarily disconnected, so it never arrived and the placeholder had no way to recover.

The placeholder now gives up on its own if nothing arrives in time and shows a clear "Image failed" state instead of scanning forever — so you know to simply ask Dr. Olivia for the image again rather than waiting on something that isn't coming.

Watch images render in progress New

In the text chatbot, generated images now appear progressively — instead of a blank gray box for the full minute or two it takes to render, you'll see the picture come in as a rough draft first and sharpen into the finished diagram. It's immediate confirmation that something is actually happening, and a preview of what's on the way.

This kicks in for freshly generated images; instant cache hits (when someone at your school already asked for something similar) still appear right away. In the avatar, the full-screen image still arrives complete once it's ready.

Image generation no longer blocks the conversation Improved

When you ask Dr. Olivia to draw, sketch, diagram, or visualize something, she now keeps teaching while the image renders in the background. The pending placeholder appears in your bubble immediately and the finished image arrives alongside her ongoing explanation — no more long awkward silences while the image generates.

• No announcement. Dr. Olivia doesn't say "I'm putting one together for you" or "I sent you a diagram." She references the structures naturally as if the image were already on screen.
• Silent cache hits. When a similar diagram has already been generated for your institution, it's reused instantly without any caveat — just the image, no commentary.
• No more duplicates. The previous behavior could occasionally generate the same image twice if the avatar connection timed out mid-generation. That whole class of issue is gone.

See the Image Generation guide for the full breakdown of what to expect.

Quiz generation now non-blocking Improved

Practice quizzes ("quiz me on X", "test me on Y") now run on the same fire-and-forget pattern as image generation. Dr. Olivia tells you a quiz is coming and immediately bridges into a brief transition (setting up what's about to come) — she no longer pauses the conversation while the quiz is being put together. The interactive quiz overlay appears as soon as it's ready.

Cached images now show up in the session email Fix

When Dr. Olivia served a cached image (a diagram a classmate had already generated for the same prompt), it never made it into the post-session summary email. The email looked for images "generated by this request" and a cache hit doesn't generate anything. Fixed: the database now records every delivery of an image to a request, so cached and fresh images alike land in the email under the exchange that delivered them.

Images now appear inside Dr. Olivia's reply Improved

In avatar mode, generated images used to land in a separate bubble below Dr. Olivia's response, which made them feel disconnected from the explanation. They now fold INTO her reply — the same look the text chatbot has always had — so the diagram sits right beside the text that's teaching you about it.

Dr. Olivia now tees up the image she's drawing Improved

She used to teach as if the image were already on screen, which made Socratic follow-ups ("Walk me through what's happening here…") sound like they needed an immediate answer — even though you hadn't seen the image yet. She now briefly frames the question as "when the image renders you'll see X — once you can see it, walk me through…" so it's clear you should wait for the diagram before answering.

Dr. Olivia shrinks to the corner while you look at an image New

When an image opens, the avatar tile now shrinks (with a smooth animation) to a small floating window in the bottom-right of the left video panel — the transcript on the right stays untouched. You can keep talking to Dr. Olivia while interacting with the image, similar to a Zoom screen share. A small × on the avatar tile hides her entirely and the button morphs into a Show Dr. Olivia pill so you can bring her back with one click. Closing the image restores the normal layout automatically.

No more duplicate images from a single request Fix

When you paused mid-sentence while asking for an image ("Show me… a normal ECG"), Tavus's voice-activity detector occasionally split your speech into two turns, and Dr. Olivia would generate two images. The system now waits until your full thought is complete before kicking off generation, so only the final version of your request produces an image — typically the more complete one. The placeholder appears immediately as before; the only change is that one image arrives instead of two.

Dr. Olivia no longer interrupts to narrate image status Improved

She used to say "I'm putting an image together for you" the moment you asked, and then check in every minute ("still working on it", "almost there", "this is taking a little longer than expected…") while it rendered. With image generation no longer blocking the conversation, those pop-ins clobbered the actual teaching response — especially when you'd already moved on to a new topic or asked for a second image.

She now stays silent about the image's status. The pending placeholder in the chat bubble is your visual signal that something's rendering, and her teaching response (about the topic itself) plays through naturally during the wait. If a generation ultimately fails the placeholder clears with a quiet failure state — just ask again.

Avatar-mode timeout workarounds removed Internal

With both image and quiz generation off the LLM completion's critical path, the avatar-mode "clean-close before long tool" plumbing (SendDoneEarly, ReceiveAvatarEcho, ConversationReconciler, the pre-tool spoken filler injection, the SuppressFillerMessage tool flag) has been removed. The completion now flows naturally for both avatar and text chatbot.

Text Input on the Avatar New

The avatar now has a dedicated text-input mode for students in quiet environments (libraries, shared study rooms) where speaking out loud isn't practical. See the Avatar Conversation guide.

• Floating Text button. A round button in the lower-left of the avatar tile toggles typing mode.
• Auto-mute / auto-unmute. Opening text mode mutes the microphone; closing it (or unmuting via the toolbar) restores audio. Manual mutes outside text mode are left alone.
• Chat-style textarea. Auto-grows up to five lines, supports Enter-to-send / Shift+Enter for newlines / Esc to close.
• Same response pipeline. Dr. Olivia speaks the answer through the avatar and writes it to the transcript — no separate "text-only" flow.

Image Generation Fixes

• Avatar no longer reads citation markup. Fixed a bug where Dr. Olivia would vocalize the raw <=== JSON ===> separator and the JSON citation block after an image was generated. Citations now route to the materials panel as intended.
• Text chatbot images land in the right bubble. When you typed a follow-up question while an image was still rendering, the finished image used to attach itself to your newest reply instead of the placeholder. It now correctly replaces the placeholder, even if other turns happened in between.
• Reopen transcript with an image up. The collapsed-transcript reopen tab now sits above the image overlay, so you can re-expand the transcript to read what Dr. Olivia said without dismissing the image first.
• Image-prompt turns no longer disappear from the summary email. When an image generation succeeded but the avatar timed out before Dr. Olivia's text reply could be saved, the whole turn was being filtered out of the session-summary email — so the student saw the image vanish and lost the context of what they asked. Those turns are now included with the student's prompt and the generated thumbnail, even when the assistant text didn't make it to the database.

Sentiment Analysis on Sessions New

Every completed session is now scored for the student's emotional state and overall sentiment by a dedicated post-session function. Drives upcoming analytics views and helps the team spot frustration, confusion, or disengagement patterns across cohorts.

• Cross-vendor evaluation. Anthropic Claude (Haiku 4.5) does the scoring so the same vendor isn't both generating and evaluating the conversation.
• Pinned taxonomy. Fine-grained sentiment (Very Positive through Very Negative) plus emotions (Happiness, Anger, Sadness, Fear, Surprise, Disgust, Confusion, Frustration, Curiosity, Engagement). Each is scored 1-100 as the model's confidence the metric is present.
• Snapshot, not timeline. Re-running the analysis overwrites the prior snapshot — there is no history kept.

Text Chatbot Reliability

• Session-end is more reliable. The text chatbot's "I'm leaving" signal previously failed silently in cross-origin development setups (and could fail in production for students with strict third-party cookie settings). It now uses a JWT-authenticated request that survives page unload, so background processes (session summary, USMLE classification, email) trigger consistently.

Cancellation & Interrupts

• User interrupts now actually cancel the response. When you start a new question while Dr. Olivia is still answering the previous one, the in-flight response is aborted instead of running to completion in the background. Cuts wasted compute and stops orphaned partial replies from polluting the conversation history.
• Interrupted turns flagged correctly. Partial responses that get cancelled mid-stream are now marked as cancelled in the database. Downstream consumers (session summaries, quiz generation, conversation reconciliation) skip them so they don't pull in unfinished text.

Privacy & Display

• Internal control tags no longer leak into student-facing views. Fixed a bug where Tavus's audio-emotion analysis (<user_audio_analysis> and related wrappers) could appear inline in the student's session-history view as if the student had said it. Now stripped at the block level alongside <user_appearance>, <user_emotions>, and other control markup.

Admin: Sentiment Throughout

• Dashboard. New Overall Sentiment card with a 3-bucket pie (Positive / Neutral / Negative) and a weighted 1.0–5.0 mood score, plus an Emotion Averages bar chart covering the full pinned taxonomy.
• Session detail. Mood pill in the session header (colored by score) gives an at-a-glance read; clicking it opens the Session Insights flyout with the full sentiment breakdown alongside topics and usage scores.
• User detail. Mood stat card joins First Session / Last Session / Sessions / Total Time, with the weighted score and descriptor. Pie + emotion breakdown collapsed below for the full picture.

Admin: Session Insights Flyout

• Renamed and reorganized. The former "Pipeline Stats" flyout is now "Session Insights" — it hosts Session Mood, Topic Understanding pies, Usage Scores, and Request Stats in one place. Bolder section titles with teal accent dividers, full-width content, zebra-striped tables, and spacing between sections.
• Usage Scores replaces the Session Totals pie. A pie chart of independent millisecond timings never really worked — the data now renders as a sorted full-width table, easier to scan and compare.

Admin: Hide Noisy Stats

• New ChatStatDefinitions.DashboardShow flag. Per-stat opt-out controlling whether a stat is included in dashboard aggregations (pipeline-performance bar chart and metrics-over-time series). Stats hidden from the dashboard are still recorded per-request and visible in the Session Insights flyout — they're just kept out of the rollup charts where they'd distort the scale. Defaults to visible; admins toggle off the misbehaving stats via SQL.

Image Generation New

Dr. Olivia can now draw images on demand — ECG strips, anatomy sketches, flowcharts, diagrams. Just ask her to draw or sketch something; no special button is required. See the Image Generation guide.

• Pending placeholder. A CRT-scan animation appears immediately so you know the request is in flight.
• Heartbeat updates. For long generations (1–3 minutes) the avatar gives short verbal updates every 60 seconds while you wait.
• Per-institution cache. Images are cached for 180 days using semantic similarity, so when a classmate asks for the same diagram you both get an instant result. Two-tier matching keeps follow-up requests in the same session sharper. Bypass the cache by asking for a fresh or different version.
• Persistent in transcripts. Each image gets its own clickable bubble in the transcript and re-opens on click. Past sessions keep their image thumbnails.
• In email summaries. Post-session emails include thumbnails linking to full-resolution images.
• Resilience. Generations that take a while are retried automatically once on failure (announced verbally), so a brief network blip doesn't kill the request.

Act-As (Impersonation) Hardening

The Act-As feature got a major correctness pass so the real vs. effective identity is consistent across every surface. See the Act-As guide.

• Personal settings stay yours. When impersonating, the My Settings dialog now shows the admin's real settings — personal API keys are never exposed to the impersonated user.
• Admin authorization uses the real account. The Administration Dashboard, audit logs, and Act-As management are all gated on the real DimId, never the impersonated one.
• Visible badge. An "Acting As User" pill appears in the header on every client surface (Selection, Avatar, Text Chatbot) so the impersonation is never hidden — especially during screen shares.
• Soft-clear with audit trail. Clearing an override writes an Active=false row instead of deleting, preserving the full history of when impersonation started and stopped.
• Email redirect. Post-session emails for Act-As sessions go to the real admin's address, not the impersonated student's.

My Settings Redesign

The personal-settings dialog is now a single form with one Save button, matching the visual language of the pre-call media dialog. See the My Settings guide.

• Single Save All. One round-trip commits every changed field; no more per-row Save buttons.
• Toggle switches. Mic-mute and video-off use the same toggle styling as the media-device dialog.
• Default Mic Mute / Video Off. Two new boolean settings pre-check the corresponding toggles in the pre-call media dialog so audio-only-by-default users don't have to flip them every time.
• Admin-only Act-As section. Visible to admins only, distinct red border so it's never confused with a personal preference.

Canvas Tooling Expansion

Dr. Olivia can now drill all the way into Canvas course content, not just list modules and assignments.

• CourseModuleItems tool. Drills into a specific module to list its items (pages, assignments, quizzes, files).
• CoursePage tool. Fetches the full body of a Canvas wiki page by URL slug.
• Module ID surfaced. The modules-and-assignments tool now includes Module IDs so chained tool calls can target a specific module.
• Data-source glossary. The system prompt now defines Canvas, ExamSoft/TPV, PubMed, and Course Materials so Dr. Olivia routes Canvas-specific questions through the Canvas API instead of guessing.

Admin Sessions Improvements

• Detail pane no longer clipped. The right-side detail pane scrolls independently — long sessions stay fully usable.
• Copy conversation button. Each conversation accordion has a Copy button that puts a markdown-formatted version of the entire transcript on your clipboard.
• Image thumbnails. Generated images appear inline in the conversation accordion and link to the full-size view.

Performance & Architecture

• File-based prompt loading. System prompts moved out of the database into Prompts/{Methodology}/{ReplicaId}.{Methodology}.md for easier editing and version control.
• Instant cache refresh. Prompt and tool reloads now propagate across the cluster within a fraction of a second — no more polling, no restarts.
• Tool-aware first-response timing. The TTFR metric now subtracts time spent waiting on tool calls in text mode, so latency dashboards reflect the actual model response time.

Responsive Header

On screens narrower than 1400 px the Selection-page header pills (Faculty Mode, Acting As User, Administration Dashboard, My Settings) collapse to icon-only badges so the header stays on a single row. Hover any icon to see its full label as a native tooltip.

Documentation

A complete in-app guide is now available at /help/. Every screen has its own page covering the buttons, badges, and workflows you'll encounter.