Overview

LitLaunch is a runtime-governance and operational-launch layer for Streamlit applications. It helps developers run Streamlit apps cleanly, predictably, and locally first, without replacing Streamlit or hiding runtime ownership.

It starts Streamlit backends, resolves browser launch strategy, supports managed browser-window and Chromium app-mode flows, provides opt-in graceful shutdown, and exposes diagnostics without pretending to secure the app itself.

LitLaunch is developed and maintained by LatticeFoundry, a software division of Sierra Cognitive Group, LLC. LitLaunch is not affiliated with Streamlit.

Why LitLaunch?

LitLaunch gets Streamlit apps running cleanly with a small amount of code or a short CLI command, while handling the operational details most projects end up reinventing.

Launch your app with sensible defaults

litlaunch app.py

CLI examples use the installed litlaunch command. python -m litlaunch ... is equivalent when a source checkout or environment has not exposed the console script on PATH.

from litlaunch import LauncherConfig, StreamlitLauncher

launcher = StreamlitLauncher(LauncherConfig("app.py"))
session = launcher.start()

That is enough to get local-first defaults, explicit backend ownership, health checks, automatic browser launch, browser capability detection, managed browser-window lifecycle where supported, clean shutdown handling, app-mode support where available, and diagnostics/reporting tools.

No shell scripts. No browser automation hacks. No custom runtime glue.

Go further with profiles and shortcuts

Profiles make launch behavior repeatable for local tools, internal dashboards, AI apps, and developer utilities:

[profiles.my-dashboard]
app_path = "app.py"
trust_mode = "strict_local"
mode = "webapp"
browser = "edge"

litlaunch create profile
litlaunch create shortcut --profile my-dashboard
litlaunch --profile my-dashboard

Simple workflows stay simple. Advanced launch, browser, monitoring, network, and diagnostics settings are there when a project needs them.

Built for real packaged applications

LitLaunch works cleanly inside packaged Streamlit applications and cross-platform runtime workflows, including apps built with tools such as PyInstaller, Inno Setup, or other desktop/runtime distribution systems.

It does not package your app or replace your installer. It handles the runtime layer those packaged apps still need: browser and app-mode startup, owned backend lifecycle, graceful shutdown, diagnostics/reporting, working-directory handling, repeatable launch workflows, and platform-specific launch behavior.

LitLaunch is designed to behave consistently across Windows, Linux, and macOS. Windows and Linux receive first-party manual validation; macOS support has limited validation until community testing broadens.

That makes it a strong fit for internal business tools, analyst dashboards, local AI utilities, and desktop-style Streamlit apps that need more operational discipline than a bare streamlit run command.

Add shutdown hooks when cleanup matters

Apps can opt into graceful cleanup without adopting a complicated application structure:

from litlaunch import LauncherRuntime

runtime = LauncherRuntime.from_env()


@runtime.shutdown_hook(label="Saving app state")
def save_state():
    ...


runtime.enable_shutdown_endpoint()

Shutdown hooks are useful for saving state, syncing data, local AI workflows, temporary resource cleanup, logging/export tasks, and other app-owned cleanup. They are optional when you need them and invisible when you do not. Routine hook messages can be marked verbose-only, and important success messages can opt into quiet mode with show_in_quiet=True; failures remain visible in normal and quiet output. Hooks that need run-specific messages can return ShutdownHookStatus so app cleanup still flows through LitLaunch's standard Hook: console lines instead of raw app prints.

Runtime governance without enterprise bloat

LitLaunch helps teams run Streamlit apps more safely and predictably without claiming to secure Streamlit applications.

It provides localhost-first defaults, trust modes, intentional network exposure workflows, runtime exposure diagnostics, Streamlit-native TLS awareness, governance/posture reporting, and sanitized diagnostics bundles. Normal localhost workflows stay frictionless; advanced runtime controls are available when the app is intentionally exposed beyond the local machine.

Generate app-owned support pages

LitLaunch can generate a Streamlit-native diagnostics/support page that host apps own, mount, and customize themselves. It gives packaged apps, internal tools, and local-first products a ready support hub for runtime posture, diagnostic sections, support artifacts, operational snapshot charts, and an optional runtime event trail.

from litlaunch import create_diagnostics_page

create_diagnostics_page(
    output_path="ui/litlaunch_diagnostics.py",
    app_name="My App",
    profile_name="my-webapp",
)

Then mount the generated function wherever it belongs in your Streamlit app:

from ui.litlaunch_diagnostics import render_litlaunch_diagnostics

render_litlaunch_diagnostics()

The generated page imports Streamlit inside app-owned generated code; LitLaunch itself does not add Streamlit as a package dependency. theme="auto" is the default and preferred mode, dark is the polished LitLaunch/RoleThread support style, and light is a functional starting point for light apps. The generated theme tokens are centralized so teams can quickly adjust the page to match their product.

What It Solves

• Start Streamlit through explicit, shell-free command construction.
• Own and stop only the Streamlit backend process LitLaunch starts.
• Open a managed browser window or Chromium app-mode window.
• Resolve Edge, Chrome/Chromium, and default-browser capability.
• Provide tokened loopback graceful shutdown hooks for app cleanup.
• Inspect local runtime readiness without launching the app.
• Keep failure output calm, concise, and actionable.

Runtime Philosophy

LitLaunch is infrastructure, not magic orchestration.

• Backend ownership is explicit through RuntimeSession.
• Browser processes are launched but never owned, killed, or controlled.
• Window monitoring is observational only. LitLaunch may observe a managed browser/app window, but it does not kill or close browser processes.
• Commands are argument tuples, never shell strings.
• Runtime dependencies remain stdlib-first; Python 3.10 uses the lightweight tomli backport for TOML profile loading.
• Diagnostics are sanitized and avoid raw environment dumps.
• Localhost is the default. Non-loopback host bindings require explicit acknowledgement because LitLaunch does not secure Streamlit itself.
• Trust modes declare operational intent: development, strict_local, or internal_network.
• Diagnostics report runtime governance, runtime exposure, and transport security posture, including Streamlit-native TLS awareness.

See docs/philosophy.md and docs/architecture.md for the full ownership model, and docs/security.md for trust boundaries.

Install

From a source checkout:

python -m pip install -e .[dev]

After changing versions or build metadata in a source checkout, rerun the editable install command above. Python import metadata is produced during installation, so stale editable installs can report an older package metadata version even when litlaunch.__version__ has changed in the source tree.

After package publication, normal installs will use:

python -m pip install litlaunch

The development environment currently uses Python 3.14.5. Package metadata allows Python 3.10 and newer, and CI currently tests Python 3.10 through 3.14 across Windows, Linux, and macOS.

Quickstart

Run an app from the CLI:

litlaunch examples/minimal_app/app.py

The examples/minimal_app path exists in a source checkout. Installed package users should point LitLaunch at their own Streamlit app path. The explicit litlaunch run examples/minimal_app/app.py form remains available for scripts and power-user workflows.

Use workflow help when you want guidance instead of command reference:

litlaunch help
litlaunch help launch
litlaunch help diagnostics

Use litlaunch --help or litlaunch run --help for argparse command and flag reference.

Run in Chromium app-mode:

litlaunch examples/minimal_app/app.py --mode webapp --browser auto

Inspect local readiness without launching:

litlaunch report examples/minimal_app/app.py

Use a reusable project profile:

litlaunch create profile
litlaunch create shortcut --profile my-webapp

The profile wizard offers Simple and Advanced modes. Simple mode defaults to LitLaunch's recommended app-window experience, detects common app-root defaults such as app.py, writes litlaunch.toml, and can optionally create a project-local launch shortcut after the profile is saved. Advanced mode exposes the fuller runtime surface for network, browser, monitor, args, cwd, and env settings.

[profiles.my-webapp]
app_path = "app.py"
title = "My App"
mode = "webapp"
browser = "edge"
trust_mode = "development"
port = 8501
auto_port = false
headless = true
runtime_event_log = ".litlaunch/runtime-events.log"
graceful_timeout = 15

[profiles.my-webapp.window_monitor]
enabled = true
appear_timeout = 60
poll_interval = 1
stable_polls = 2

litlaunch --profile my-webapp
litlaunch command --profile my-webapp
litlaunch report --profile my-webapp

Profiles can live in litlaunch.toml or under [tool.litlaunch] in pyproject.toml. Explicit CLI flags override profile values. Bare profile names such as litlaunch my-webapp are intentionally not launch shorthand; use --profile so profile-based launches stay unambiguous.

litlaunch report is the recommended human-facing diagnostics workflow. It writes .litlaunch/reports/litlaunch-report.html by default. The explicit litlaunch inspect --html, --json, and --bundle commands remain available for power-user and machine-readable diagnostics.

Generated project artifacts

LitLaunch keeps generated files under a project-local .litlaunch/ directory by default:

.litlaunch/
  reports/              HTML reports, JSON output, and support bundles
  shortcuts/            generated launch shortcuts
  tmp/browser-profiles/ managed temporary Chromium profiles

Keep litlaunch.toml in the project root when you want profiles to travel with the app. Add .litlaunch/ to .gitignore when generated reports, shortcuts, and runtime scratch files should stay out of source control. Explicit --output paths still write exactly where you ask.

Shortcut generation uses native project-local artifacts by default: .lnk on Windows, .desktop on Linux, and a small .app bundle on macOS. Use litlaunch create shortcut --profile NAME --kind script when you prefer the simple .bat, .sh, or .command script form. macOS shortcut support has limited validation until broader community testing expands.

Python integrations can run the same configured profile through run_profile():

from litlaunch import load_profile, run_profile

profile = load_profile("my-webapp")
result = run_profile(profile)

if result.exit_code:
    print(result.message)

If profile.window_monitor is enabled, run_profile() uses the monitored webapp runner. Otherwise it starts the normal StreamlitLauncher path and returns the same structured MonitoredRunResult shape.

Use the Python API:

from litlaunch import LauncherConfig, StreamlitLauncher

config = LauncherConfig(
    app_path="app.py",
    title="My Streamlit App",
    mode="browser",
)

session = StreamlitLauncher(config).run()

try:
    print(f"Running at {session.url}")
finally:
    session.stop()

run() is the friendly common entry point. start() is the explicit lifecycle entry point; both return a RuntimeSession.

For app-mode integrations that need close-to-shutdown behavior, LitLaunch also provides a high-level monitored runner:

from litlaunch import LauncherConfig, LaunchMode, run_monitored_webapp

result = run_monitored_webapp(
    LauncherConfig(
        app_path="app.py",
        title="My Streamlit App",
        mode=LaunchMode.WEBAPP,
        browser="edge",
    ),
    graceful_timeout_seconds=15,
)

The monitored runner observes the app window and returns a MonitoredRunResult. It does not own, kill, close, or control browser windows.

Show launch behavior without starting Streamlit or opening a browser:

plan = StreamlitLauncher(config).build_launch_plan()
print(plan.command_display)
print(plan.app_url)

build_launch_plan() is useful for diagnostics, integration tests, and configuration parity checks. Sensitive command and environment values are redacted in display fields.

Advanced integrations can inject a backend command provider when a packaged or embedded app needs a custom backend executable while LitLaunch still owns env injection, health checks, browser launch, and RuntimeSession lifecycle:

from litlaunch import BackendCommand, LauncherConfig, StreamlitLauncher


class PackagedBackend:
    def build_backend_command(self, context):
        return BackendCommand(
            (
                "dist/MyApp/MyApp.exe",
                "--host",
                context.host,
                "--port",
                str(context.port),
            ),
            description="packaged Streamlit backend",
            backend_kind="packaged",
        )


launcher = StreamlitLauncher(
    LauncherConfig(app_path="app.py"),
    backend_command_provider=PackagedBackend(),
)

Custom backend commands must bind the requested host/port and expose the Streamlit health endpoint used by LitLaunch.

Public API Surface

The supported public surfaces are the configuration, launcher, launch-plan, profile, monitored-runner, backend-command-provider, app shutdown, and inspect diagnostics APIs documented in docs/architecture.md. Window provider internals, low-level browser/window matching details, and console presentation internals are implementation details and may evolve faster than the public API.

Supported Capabilities

Common CLI Examples

litlaunch version
litlaunch platform
litlaunch browsers

litlaunch command app.py --server.runOnSave true -- --workspace demo
litlaunch run app.py --mode browser
litlaunch run app.py --mode webapp --browser edge
litlaunch run app.py --port 8501 --no-auto-port
litlaunch run app.py --host 0.0.0.0 --allow-network-exposure
litlaunch run app.py --trust-mode strict_local
litlaunch run app.py --host 0.0.0.0 --trust-mode internal_network --allow-network-exposure
litlaunch run app.py --mode webapp --monitor-window --title "My Streamlit App"
litlaunch run app.py --mode webapp --monitor-window --graceful-timeout 15
litlaunch run app.py --mode webapp --monitor-window --monitor-appear-timeout 90

litlaunch inspect
litlaunch inspect app.py --html --output litlaunch-report.html
litlaunch inspect app.py --json
litlaunch inspect app.py --bundle --output litlaunch-report.txt --force
litlaunch report app.py --host 0.0.0.0 --trust-mode internal_network --allow-network-exposure

Unknown arguments before Streamlit's -- separator are passed through to Streamlit. Arguments after -- are passed to the app.

Python integrations can set LauncherConfig.cwd and LauncherConfig.extra_env to control the backend child process without mutating global environment state.

App-Mode And Monitoring

Browser mode can use a managed temporary Chromium profile and a new top-level browser window so closing that window can trigger graceful shutdown where supported. Webapp mode opens a Chromium app-mode window when an app-mode capable browser is available. Monitoring remains observational: LitLaunch does not own, close, or kill browser processes. If browser-window monitoring cannot identify a window confidently, Ctrl+C remains the shutdown path.

See docs/browser_support.md and docs/window_monitoring.md.

Inspect And Troubleshooting

Use litlaunch inspect before launching or when a runtime fails. Plain inspect prints format guidance; HTML is the recommended human-readable diagnostics report, JSON is for tools, and bundle output is for support:

litlaunch inspect app.py --html --output litlaunch-report.html
litlaunch inspect app.py --json --output litlaunch-report.json
litlaunch inspect app.py --bundle --output litlaunch-report.txt

Reports include runtime governance, runtime exposure, and transport security sections. They show trust mode, host exposure scope, acknowledgement state, Streamlit-native TLS posture, and plaintext network-exposure risk. This is operational visibility, not a security guarantee: LitLaunch does not authenticate users, terminate TLS, or secure Streamlit applications.

See docs/inspect.md and docs/troubleshooting.md.

Documentation

• Overview
• Philosophy
• Installation
• Quickstart
• CLI
• Security And Trust Boundaries
• Runtime Events
• Diagnostics Page Generator
• Browser Support
• Window Monitoring
• Inspect
• Troubleshooting
• Architecture
• RoleThread Lite Integration
• Packaging Notes

Non-Goals

• Owning, killing, or controlling browser processes.
• Killing by process name, PID discovery, or port owner.
• Replacing Streamlit's CLI/config system.
• Running a local diagnostics dashboard or diagnostics server today.
• Owning PyInstaller, Nuitka, installer, updater, or full packaging workflows. LitLaunch does support project-local native shortcut generation for existing profiles, with script fallback when needed.
• Adding terminal UI frameworks or heavy runtime dependencies.

Versioning

LitLaunch uses 0.0.0 style internal versioning:

• Patch bumps are for fixes, cleanup, and basic hardening passes.
• Minor bumps are for larger internal milestones and feature work.
• Major versions are controlled manually by the project owner.

License

MIT. Copyright (c) 2026 Sierra Cognitive Group, LLC.