mts1b-frontends

All frontends: webui (Next.js), TUI (Textual), CLI (mts + Python helpers).

Repo: github.com/MTS1B/mts1b-frontends Layer: 7 Wave: 2 (months 4-7) Depends on: foundation, every service repo's gRPC/REST API Audience: every operator

What it is

Three UIs sharing a common API client + visualization layer:

Frontend	Audience	When to use
webui (Next.js)	operators, researchers	day-to-day desk view, dashboards, deep dives
TUI (Textual)	over SSH, ops	when you only have a terminal
CLI (`mts`)	scripts, automation	composable, pipeable, scriptable

Module layout

mts1b_frontends/
├── webui/
│   ├── apps/web/                # Next.js 14 app
│   │   ├── app/
│   │   │   ├── desks/<desk>/page.tsx
│   │   │   ├── replay/page.tsx  # backtest replay UI
│   │   │   ├── matrix/page.tsx  # session matrix
│   │   │   ├── lab/page.tsx
│   │   │   ├── signals/page.tsx
│   │   │   ├── zilla/page.tsx
│   │   │   └── ...
│   │   └── lib/api/             # auto-gen from foundation OpenAPI
├── tui/
│   ├── apps/textual/            # Textual app
│   │   ├── screens/
│   │   │   ├── positions.py
│   │   │   ├── orders.py
│   │   │   ├── halts.py
│   │   │   └── ...
│   │   └── widgets/
├── cli/
│   ├── bin/mts                  # entrypoint
│   ├── cmds/
│   │   ├── cmd_halt.py
│   │   ├── cmd_resume.py
│   │   ├── cmd_cancel_all.py
│   │   ├── cmd_flatten_paper.py
│   │   └── ...
│   └── helpers/                 # python helpers for scripts
└── api_client/                   # shared client used by all three
    ├── __init__.py
    ├── client.py
    └── generated/                # auto-gen from OpenAPI specs

webui (Next.js)

Multiple "desks" — focused views per workflow:

Desk	Purpose
Replay	step-through backtest with bookmarks + screenshots + manual paper-trade
Matrix	session-matrix grid (6 sessions × 5 asset classes) of candidate strategies
LAB	parameter sweep visualization, sensitivity heatmaps, MC/bootstrap CIs
Signals	cross-asset signal IC matrix, decay curves, conditional distributions
Zilla	40+ pre-built and adaptive baskets vs SPY
Funds	per-fund NAV, exposure, P/L attribution
Risk	live envelope status, gate-failure breakdown
Ops	halt control, audit timeline, watchdog status

Hot-reload during dev: npm run dev in webui/apps/web.

TUI (Textual)

Same primitives as webui but rendered for the terminal. Useful over SSH or during a panic:

mts tui

Keyboard navigation, mouse where supported. All operator commands (halt / cancel-all / flatten / resume) are one keystroke.

CLI (`mts`)

The most composable interface. Designed for pipes + scripts:

mts mts1b-treasury funds list --json | jq '.[] | select(.broker == "ibkr")'

mts mts1b-oms positions list --fund-id live-momentum --format csv > positions.csv

mts mts1b-research strategies list --enabled-only \
  | mts mts1b-research strategy show --strategy-id $(awk '{print $1}')

Top-level commands by service:

mts mts1b-treasury        funds, transfers, nav
mts mts1b-oms             orders, positions, fills
mts mts1b-research        strategies, factors, ladder, drift
mts mts1b-riskengine      envelope, gates, halts
mts mts1b-brokers         status, test
mts mts1b-deploy          install, status, logs, restart, ...
mts mts1b-platform        tail (NATS), audit
mts cmd                   halt, resume, cancel-all, flatten-paper

Plus shorthand:

mts ls                   # positions across all funds
mts halt                 # firm-wide halt with confirmation
mts logs <service>       # tail a service's logs

API client (shared)

Auto-generated from each service's OpenAPI spec (which is derived from mts1b-foundation types):

// webui/apps/web/lib/api/oms.ts (generated)
import { OmsClient } from "@mts1b/api-client";

const oms = new OmsClient({ baseUrl: "http://localhost:8001" });
const positions = await oms.getPositions({ fundId: "paper-momentum" });

# tui + cli use the same client (Python)
from mts1b_frontends.api_client import OmsClient

oms = OmsClient(base_url="http://localhost:8001")
positions = await oms.get_positions(fund_id="paper-momentum")

Regenerated whenever mts1b-foundation ships a new minor version.

Auth

OAuth2 (Authentik) for webui; static token for TUI/CLI (per-operator). Tokens stored in ~/.mts1b/auth.json (chmod 600).

mts auth login                  # opens browser to Authentik
mts auth whoami
mts auth logout

Themes

Dark + light. webui is dark by default. TUI follows terminal color scheme. CLI uses Rich for formatting (auto-detects 256-color / truecolor / no-color).

Build + run

# webui
cd webui/apps/web
npm install
npm run dev                    # localhost:3000

# TUI
cd tui/apps/textual
pip install -e .
mts tui

# CLI
cd cli
pip install -e .
mts --help

Deployment

Via mts1b-deploy:

webui: containerized Next.js standalone build, served by Caddy
TUI: not deployed; users install locally for SSH connections to the host
CLI: pre-installed in every Docker image; users pip install mts1b-cli on their workstation

Roadmap

Version	Items
0.1 (Wave 2)	webui (replay/matrix/lab/signals/zilla/funds/risk/ops), TUI, CLI
0.2 (Wave 2)	mobile-responsive webui, Authentik OAuth
0.3 (Wave 3)	Real-time WebSocket updates across the board
0.4 (Wave 3)	iPad/touch optimization
1.0 (LTS)	Stable URL structure

What it is​

Module layout​

webui (Next.js)​

TUI (Textual)​

CLI (mts)​

API client (shared)​

Auth​

Themes​

Build + run​

Deployment​

Roadmap​

See also​