fast-dash

Fast Dash

Fast Dash turns a Python function into a Plotly Dash web app. The @fastdash decorator reads the function's signature, maps each parameter's type hint to a UI component, maps the return type to an output component, and serves the result. No frontend, no callbacks, no boilerplate.

When to use this skill

Use this skill when the user:

• has a Python function and wants a UI for it ("make this a web app", "add a form", "dashboard around this")
• is prototyping an ML / data / API tool and wants shareable interactivity
• needs cascading inputs, a multi-step wizard, or multiple tools in one app
• wants an AI agent (Claude Code, Cursor, ...) to drive the app over MCP, or a chat assistant beside it

Do not use this for: production apps with complex routing, custom auth, or non-Python frontends — Fast Dash is opinionated for the single-file-Python-function use case.

Install

pip install fast-dash

The core pattern

from fast_dash import fastdash

@fastdash
def greet(name: str = "world") -> str:
    return f"Hello, {name}!"

# Serving on http://127.0.0.1:8080

That is the whole app. Open the URL, type a name, click Run.

How to approach a Fast Dash task

1. Start from the user's function. If they don't have one, write the smallest function that captures their intent.
2. Add type hints and defaults. This is where the UI comes from — int → number input, bool → checkbox, str with a list default → dropdown, etc. See references/components.md for the full table.
3. Decorate with @fastdash. For more control (tabbed multi-function apps, multi-step pipelines), use the FastDash(...) class directly.
4. Run and verify. The decorator starts a server immediately on import. For notebooks, pass mode="inline".

Common patterns

Pattern	Syntax	When
Single function	@fastdash	One tool, one form
Multiple outputs	-> (Graph, Graph) + mosaic="AB"	Dashboard with several plots
Cascading inputs	state=depends_on("country", resolver)	Dependent dropdowns
Multiple tools, one app	FastDash([fn_a, fn_b], tab_titles=[...])	Tabbed "apps" under one URL
Multi-step wizard	FastDash(steps=[fn_a, fn_b, fn_c]) + from_step(prev_fn)	Pipeline UX, one panel at a time
Streaming outputs	yield partial results (or update("x", chunk)) + stream=True	LLM / token-by-token / progress
Notebook rendering	@fastdash(mode="inline")	Jupyter
Wrap a custom component	Fastify(dcc.Slider(...), "value")	Any Dash component
Drive from an AI agent	@fastdash(mcp_server=True) → agent connects to :8080/mcp	Human and agent use the same app
Agent-built form	DynamicDash(callback_fn=..., mcp_server=True) + agent calls set_form	Form fields decided at runtime
Chat assistant	@fastdash(chat=True) on a yield-ing function, or chat=True, chat_model=... on a normal app	Chat UI / assistant beside an app