Review fixes for the Cron Recipes stack before release: - hydration-move: */90 in the cron minute field silently wraps to hourly (croniter-verified) — 90/120-minute options never fired at their stated cadence. Replaced with an hour-field step (0 9-17/2 * * 1-5) and an interval_hours slot whose options (1/2/3h) all fire as labeled. - fill_recipe: reject unknown slot names. A typo'd 'tiem=07:15' used to silently create the job at the 08:00 default; now it 422s on the dashboard form and errors on the slash/deep-link paths with the valid slot list. - deliver slot: non-strict enum (options are suggestions, scheduler validates downstream) so slack/whatsapp/etc. users aren't locked out; GET /api/cron/recipes rewrites its options from cron_delivery_targets() so the dashboard form only offers configured platforms; help text no longer claims dashboard-created jobs deliver to 'the chat you set this up from' (the endpoint strips origin — they go to the home channel). - gateway: success/accept messages no longer point at /cron (cli_only); surface-aware hint instead. Conversational fill now sends the 'Setting up X — I'll ask you a couple of things…' ack before the agent turn, matching the CLI experience. - important-mail catalog entry: reference the urgency classifier by module path (python3 -m cron.scripts.classify_items) instead of baking an absolute host path into the job prompt — stale after relocation and nonexistent on remote terminal backends. cron/scripts is now a real package and ships in the wheel (pyproject packages.find). - export_recipe: interval schedules round-trip again — parse_schedule stores 'minutes' but the renderer only read 'seconds', so every interval job exported as the silent '0 9 * * *' fallback. - skills_hub install: say so when a recipe suggestion is dropped (latched dedup or pending cap) instead of printing nothing. Targeted tests: 58 cron/recipe + 261 web_server pass; E2E-validated all 14 recipes fill+parse, hydration cadences via croniter, typo rejection on slash + endpoint paths, surface-aware hints, and interval export round-trip. |
||
|---|---|---|
| .. | ||
| public | ||
| src | ||
| eslint.config.js | ||
| index.html | ||
| package.json | ||
| README.md | ||
| tsconfig.app.json | ||
| tsconfig.json | ||
| tsconfig.node.json | ||
| vite.config.ts | ||
Hermes Agent — Web UI
Browser-based dashboard for managing Hermes Agent configuration, API keys, and monitoring active sessions.
Stack
- Vite + React 19 + TypeScript
- Tailwind CSS v4 with custom dark theme
- shadcn/ui-style components (hand-rolled, no CLI dependency)
Development
# Start the backend API server
cd ../
python -m hermes_cli.main web --no-open
# In another terminal, start the Vite dev server (with HMR + API proxy)
cd web/
npm install
npm run dev
Open the Vite URL printed in the terminal (usually http://localhost:5173). That is the live-reload UI.
hermes dashboard on port 9119 serves the built bundle from hermes_cli/web_dist/, not the Vite dev server — changes in web/src/ will not appear there until you run npm run build and restart the dashboard (or use web --no-open + Vite as above).
The Vite dev server proxies /api requests to http://127.0.0.1:9119 (the FastAPI backend).
Build
npm run build
This outputs to ../hermes_cli/web_dist/, which the FastAPI server serves as a static SPA. The built assets are included in the Python package via pyproject.toml package-data.
Structure
src/
├── components/ui/ # Reusable UI primitives (Card, Badge, Button, Input, etc.)
├── lib/
│ ├── api.ts # API client — typed fetch wrappers for all backend endpoints
│ └── utils.ts # cn() helper for Tailwind class merging
├── pages/
│ ├── StatusPage # Agent status, active/recent sessions
│ ├── ConfigPage # Dynamic config editor (reads schema from backend)
│ └── EnvPage # API key management with save/clear
├── App.tsx # Main layout and navigation
├── main.tsx # React entry point
└── index.css # Tailwind imports and theme variables
Typography & contrast rules
Read before adding or editing UI styles. These rules keep the dashboard legible across all built-in themes and stop drift back into the patterns the design system was just refactored out of.
Text size floor
- Minimum body size:
text-xs(12px / 0.75rem). Do not use arbitrarytext-[0.6rem],text-[0.65rem],text-[9px],text-[10px], ortext-[11px]on copy, hints, labels, counts, or badges. Use the standard scale:text-xs,text-sm,text-base. - Smaller sizes are only acceptable on decorative overlays (chart stripes, empty-state icons) — never on text the user is meant to read.
Opacity floor on text
- Never apply opacity below 0.7 to text. No
opacity-30,opacity-50,opacity-60on<span>s,<p>s, labels, etc. - Do not stack opacity tokens. Patterns like
text-muted-foreground/60,text-midground/70,text-foreground/50create unpredictable WCAG failures because the parent token already has alpha. - Use the semantic text tokens from
@nous-research/ui'sglobals.css:text-text-primary— default body text.text-text-secondary— subtitles, meta, inactive nav.text-text-tertiary— small chrome labels, counts, footnotes.text-text-disabled— disabled states.text-text-on-accent— text on filled accent surfaces.
Brand uppercase via text-display, not raw uppercase
- The dashboard preserves the Nous brand uppercase aesthetic, but it is opt-in per element, not global.
- Apply uppercase via the DS utility
text-displayon brand chrome only — page titles, nav section headings, badges, brand wordmark. DS components (Button,Badge,Tabs,Segmented, etc.) already self-applytext-display. - Do not introduce new
uppercase(the literal Tailwind class) inhermes-agent/web/src. Prefertext-displayfor new brand chrome. Legacyuppercasecall sites (e.g.components/ui/label.tsx,card.tsx) remain until migrated. - The app shell no longer forces uppercase globally, so blanket
normal-caseopt-outs are unnecessary. Usenormal-caseonly where a DS component appliestext-displaybut the label should stay sentence case — e.g. dynamic user content (model slugs, theme names) or fixed UI copy that is not brand chrome (EnvPage “not configured” toggle, sidebar “New chat”).
Fonts
Typography is opt-in per surface, not global on layout shells — the app shell and page header keep their original theme/expanded fonts; Mondwest applies only where explicitly set.
| Tier | Classes | Use for |
|---|---|---|
| Brand chrome | font-mondwest text-display (or themedChrome) |
Sidebar nav, card section headers (CardTitle), Segmented filter buttons, filter panel headings |
| Themed body | font-mondwest normal-case (or themedBody) |
Card content (Card, CardDescription), session/platform rows, analytics tables — scoped to the component |
| Page chrome | font-expanded |
Page header h1 (PageHeaderProvider) — sentence case, not text-display |
| Wordmark | Typography + size/tracking only |
Sidebar/mobile “Hermes Agent” — mixed case, no Mondwest, no text-display |
| Technical | font-mono-ui / font-mono / font-courier |
Model slugs, env keys, schedules, YAML, repo URLs |
- Do not put
themedBodyorthemedFonton<main>,App, or other layout wrappers — it overrides component-scoped styles. CardappliesthemedBody;CardTitleusestext-display(uppercase chrome);CardDescriptionusesthemedBody.NouiTypographydefaults tofont-sansunless a font prop is passed.- Do not use raw
font-sansorfont-display(theme sans variable) on new dashboard UI — prefer Mondwest tiers above where brand-appropriate.
Color tokens
- Prefer semantic tokens (
text-text-*,bg-card,border-border,text-foreground,text-destructive,text-success,text-warning) over raw layer references (text-midground,text-foreground). text-muted-foregroundis now wired to--color-text-secondary, so existing call sites stay correct, but new code should prefer the semantic name.- When you genuinely need a non-token color (icon de-emphasis on a chart, terminal foreground via inline style), keep alpha at
≥ 0.7for any text.