Two operator-felt papercuts on the editor:
* Models only appeared after a manual Refresh click. Now the
editor calls `_refreshModels()` from initState (post-frame)
so chips populate immediately. Provider switch also re-fires
the refresh because endpoint / model set changes too.
* Operator had no signal which model is suitable. Tiny models
fail at instruction following; huge models stall on a
laptop. Adds a heuristic suitability rating (parses
parameter count from common names like `gemma3:4b`,
`llama3.2:70b`; falls back to family hints for OpenAI /
Claude). Each chip now carries:
- a coloured leading dot (green=balanced, orange=small
or large, red=huge, grey=unknown)
- a star icon for the hand-curated "tested with F∆I"
allowlist (gemma3:4b, llama3.2:3b, qwen2.5-coder:7b,
gpt-4o-mini, claude-haiku-4-5, …)
- a tooltip ("balanced", "huge — likely too slow on a
laptop", …)
Sorting also moves recommended → balanced → others to the
front so the operator's eye lands on green first.
* Compact legend below the chip wrap explains the colour
code in place — no lookup needed.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Three operator-facing improvements driven by Stefan's testing:
* **Test Connection** runs against the form's *current* values
and no longer requires saving first. Previously we save-then-
test which caused a "system AI is disabled" message when the
form's model field was still empty. Now the test prompt
travels with the request body so unsaved tweaks can be
validated.
* **Model picker** replaces the free-text Model field. A
Refresh button calls HubAdmin.ListSystemAiModels (provider's
`/v1/models`); installed models show up as clickable chips
that fill the field with one tap. The text input stays for
cases where the operator already knows the id.
* **Pull** button (Ollama only) downloads a model via
`/api/pull`. Type `gemma3:4b`, click Pull, spinner, then the
new model lights up in the chip list. No more "switch to a
terminal to ollama-pull" detour.
Helper text updates for the empty-model case ("Required: pick
from the list (Refresh) or type one. Use Pull to download from
Ollama.") so a fresh operator never wonders why Save / Test
errors out.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Replaces the read-only System AI block in Settings with a real
editor. Operator never has to touch ~/.fai/config.yaml again
to change provider, endpoint, model, API-key env-var, or
privacy mode.
UX choices that align with the zero-learning-curve rule:
* Provider dropdown lists Ollama / OpenAI / LM Studio / vLLM
/ Custom by friendly name. Each selection auto-fills the
endpoint and api_key_env defaults — but only when the
field is still empty or matches a different preset's
default, so manual overrides are never clobbered.
* Each provider has a one-line description rendered under
the dropdown ("Ollama → Local `ollama serve`. Models stay
on your machine. No API key needed.") and a model-hint
placeholder ("gemma3:4b · llama3.2:3b · qwen2.5-coder:7b")
that goes away when a model is typed.
* Privacy mode is a vertical radio group with each option's
description in place — no doc-lookup needed.
* "Test connection" sends an `ok`-ping and renders the
same error-kind → fix-hint mapping the audit drill-down
uses.
* "Save" persists + hot-reloads the hub. No daemon restart
required. UI status badge flips to "enabled · <privacy>"
immediately.
`fai_dart_sdk` 0.5.0 carries the underlying `updateSystemAi` /
`testSystemAi` RPCs.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Two new operator affordances powered by `HubAdmin.AskAi`:
* **Audit drill-down → "Explain" button.** Tappable on every
failed event when System AI is configured; greyed out with
a "Configure in Settings" tooltip otherwise. Shows an
inline panel under the event fields with the LLM's plain-
language explanation, a "FIX" suggestion, the privacy-mode
badge (`redacted` / `full`), and round-trip latency. Errors
map to actionable fix hints (`env_missing` → "set $VAR in
your shell, then fai daemon restart"; `network` → "is your
provider running?"; etc.) — no raw HTTP errors surfaced.
* **Settings dialog → System AI panel.** Renders the live
HubAdmin.SystemAiStatus (provider, endpoint, model,
api_key_env name, privacy mode). When disabled, shows a
one-line explainer + pointer to docs/architecture/system-ai.md.
Read-only for now; editing the config still requires
~/.fai/config.yaml (Phase-1+ adds in-place editing).
Both surfaces follow the new zero-learning-curve UX rule:
the user never has to look up what the configurable means or
how to fix an error — every state self-explains in place.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Turns Studio from read-only dashboard into operator console.
Pieces:
* **New Store page** (sidebar destination #3): browses the
hub's bundled store-index with query + category + status
filters. Each card has an Install button that prompts for
the `.fai` bundle source (URL or local path), ships it to
HubAdmin.InstallModule, shows success / error in a progress
dialog, refreshes the list.
* **Audit drill-down**: every event row is now tappable;
opens a modal with all LoggedEvent fields including the
new `detail` JSON pretty-printed in a code block. KRITIS
forensic story: every audit row → full structured detail
in two clicks.
* **Approvals payload preview**: shipped already; now
pretty-prints JSON when the preview parses as such, plus
a clearer `PAYLOAD PREVIEW` label. Reviewer identity
defaults to `$USER@studio` instead of the hard-coded
`studio-mvp` so the audit trail records who acted.
* **Channel switcher in Settings dialog**: the dialog now
pulls HubAdmin.ChannelStatus and renders one row per
channel (local / dev / beta / production) with port,
running indicator, active marker. A "Connect" button
on a running channel sets Studio's endpoint to that
channel's port and reconnects in one click.
Cmd+1..6 keyboard shortcuts updated for the six destinations.
Widget test asserts every destination renders.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Two issues from live feedback:
1. The version label said "v0.7.1 · warm-minimalism". The
suffix is an internal design-direction codename, not
user-facing. Drop it — the line now just reads "v0.7.2".
2. The ∆ pulse in live mode used only alpha + blur, which is
too subtle to register at first glance. Replace with a real
scale-pulse: the whole mark grows from 0.85x → 1.15x → 0.85x
over 1.6s. Glow halo blur range expanded (8→22) and alpha
range widened. The triangle also gets a soft fill on the
up-beat. Easy to spot from across the room.
Bumps fai_studio 0.7.1 -> 0.7.2.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Tapping a module card on the Modules page slides up a detail
sheet showing the full manifest data: directory path
(monospaced + selectable), every capability with version, and
every declared permission with a semantic icon (net = globe,
fs.read = folder, fs.write = edit, env = terminal, hub =
shield).
Empty permission list shows "(none — pure-computation module)"
explicitly — the operator should see *why* a module needed
zero perms, not be left guessing.
Backed by HubAdmin.ModuleInfo. New widget FaiModuleSheet with a
small drag-handle, FutureBuilder for the load state, error
surfacing if the RPC fails.
Bumps fai_studio 0.5.0 -> 0.6.0.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Studio is no longer hardcoded to localhost:50051. A gear icon
in the sidebar footer opens a small settings dialog where the
operator picks host / port / TLS-on-or-off. Saved values
persist via shared_preferences and Studio reconnects to the
new endpoint immediately.
- data/hub.dart: HubService.loadPersistedEndpoint() called once
before the first frame; reconnect() persists on every change.
- widgets/fai_settings_dialog.dart: typed form with port range
validation, live URL preview pill, error surfacing if the
reconnect fails.
- main.dart: gear icon at the sidebar bottom, replaces the
static "platform v0.10.42" footer.
shared_preferences added as dependency. flutter analyze clean,
flutter test 2/2, macOS debug build succeeds.
Bumps fai_studio 0.4.0 -> 0.5.0.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
Replaces the Material-default look with a deliberate visual
language. Single accent (sky-cyan), Inter + JetBrains Mono via
google_fonts, dense rows over puffy cards, micro-interactions
under 250ms. Dark-first, light reaches parity.
New foundation under lib/theme/:
- tokens.dart FaiColors / FaiSpace / FaiRadius / FaiMotion
- theme.dart ColorScheme + typography + component themes for
light and dark, plus FaiTheme.mono helper for
technical strings (IDs, paths, capability refs).
Six primitives under lib/widgets/:
- FaiCard flat card, optional accent stripe (top or
left). No shadows.
- FaiPill small inline label with five tones
(neutral / accent / success / warning /
danger), optional mono and leading icon.
- FaiStatusDot breathing dot, used as a "live" indicator.
- FaiDataRow Linear-style dense row for the audit
stream — hover-elevation, mono leading,
coloured leading stripe.
- FaiEmptyState gracious icon + title + hint + action,
replaces "(no data)" everywhere.
- FaiDeltaMark the ∆ signature element. Three modes —
idle (still), live (gentle pulse), busy
(slow rotation). Drawn from primitives,
not a font glyph. Lives in the sidebar
header so the brand is always visible.
Page-level changes:
- main.dart custom 220px sidebar replaces the Material
NavigationRail. ∆ on top, hub-connection
pill below it, hover-animated destinations,
page transitions are 200ms slide+fade.
- modules.dart FaiCard rows with capability pills.
- audit.dart FaiDataRow stream, segmented filter chips,
live status bar with hash-chain badge.
- approvals.dart FaiCard with accent-top stripe, structured
action footer, toast on approve/reject,
themed reject-reason dialog.
google_fonts added as dep. flutter analyze clean. flutter test
2/2. flutter build macos --debug succeeds.
Bumps fai_studio 0.2.0 -> 0.3.0.
Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>
