james.maesandClaude Opus 5.5
6a26a7d4fa
The demo's mock host answers "How many images do I have by type?" with count_records through the UI API and then present_results, so the panel shows the owner's example (summary, metric, table with shares and row links, pie chart, filter links, action). Screenshots light/dark at 390/1440 px with the Next UI build of qqq-frontend-next#111. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant app home and provider settings screenshots
Taken 2026-10-13 against the local AssistantDemoServer (qbit-assistant-provider-anthropic test sources), never production.
- Memory backend and the Next UI 1.0.0-RC.16 jar.
- The Anthropic API is the recorded test server: real HTTP through the real adapter, engine, Test connection and usage recording, with canned answers and no key.
Light and dark, 390 and 1440 px:
home-*: the Assistant app home.provider-*: the provider settings screen, with the masked key and the model-call child list.provider-not-set-*: a provider whose key variable is not set.provider-edit-*: the edit form.prices-*: the model price table.
0.1.0-RC.6 additions, taken 2026-10-13 with the same demo server and the Next UI 1.0.0-RC.17 jar. The demo seeds a profile ("Ada"), two skills, and data access rows (page READ_WRITE, media NONE). Its tools run through the /qqq/v1 UI API.
Light and dark, 390 and 1440 px:
home-denials-*: the home's "Recent denials" card. The analyst'scount_recordson Blog Post got HTTP 403 fromPOST /table/{tableName}/count.home-access-*: the "Data access: what the assistant can do for you" widget (Explain access), with each table's access and each tool allowed or denied, with the reason.data-access-*: the Data access table.profile-*: a profile (persona), with its personality, its scope and its Skills child list.skill-*: a skill: its instructions and when it is used.policy-*: the policy, with its "Tools and data" section.panel-persona-*: the panel with the configured persona. This is the Next UI build of qqq-frontend-next#94; the demo's admin matches the permission-scoped "Ada (analytics)" profile.panel-pending-*: the assistant proposedupdate_recordon page 12, which is READ_WRITE. The card shows the change with Save and Cancel; nothing is saved yet.panel-confirmed-*: after Save, the write ran once throughPATCH /qqq/v1/table/page/12with the user's session. The card says Done, and the record screen behind the panel shows the new value.
Planned P-a provider registry additions, taken against OpenAiDemoServer in the
OpenAI provider's test sources with the Next UI 1.0.0-RC.18 jar. See that
demo's run instructions.
Discovery and connection-test status, TEST usage and the assistant turn are real
executions against the localhost mock; replies/token counts are synthetic, not
live GPU measurements. No screenshots in this group have been added yet.
Light and dark, 390 and 1440 px:
p-provider-local-*: Local GPU (Ollama), keyless LOCAL settings, discovery/test results and discovered models.p-model-*: a discovered model's capabilities, compute prices and connection-test result.p-home-providers-*: the home's provider summary and usage from the executed demo calls, including disabled OpenRouter with its key Not set.
Answer presentation (owner priority, 2026-10-13), taken against OpenAiDemoServer with the Next UI build of QRun-IO/qqq-frontend-next#111 (next dev proxied to the demo). The demo's OpenAI-compatible mock answers "How many images do I have by type?" the way a model would through the real adapter: five count_records calls through the UI API, then present_results.
Light and dark, 390 and 1440 px:
panel-presentation-*: the summary line, an Images metric, the "Media by type" table (counts and the server-computed 80.2% and 19.8%, each row a filtered-list link), the pie chart, two filter links (Open JPEG media, Open PNG media) and a follow-up action.