Paste the URL. Upload the file. Drop in the email. Whatever form your process lives in — Word doc, PDF, wiki page, or plain text — FlowScript extracts the steps and hands you back a clean, editable flowchart. No drawing. No dragging boxes.
Vote anonymously. We're deciding what to build next.
Want to be notified when it ships? Drop your email — no spam, just a ping.
@direction TB @layout grid #start Alert Received Review monitoring dashboard #decision Severity? -> 'P1' Page On-Call Engineer -> 'P2' Assign to Team Lead -> 'P3' Add to Backlog Page On-Call Engineer Investigate & Mitigate #decision Resolved? -> 'yes' Write Post-Mortem ~> 'no' Investigate & Mitigate Write Post-Mortem #end Incident Closed
Three ways to get a flowchart. Pick the one that fits your workflow.
Got a process in a Word doc, PDF, email, wiki page, or plain text? Paste it or upload it. The generator reads whatever form it's in, extracts the process, and renders a clean diagram. One click — no drawing required.
The generated .flow opens in a Monaco-powered editor with live SVG preview. Tweak labels, add branches, fix the edges AI got wrong — syntax highlighting included.
One Docker container, your LLM key. Or skip the UI entirely — use the MCP server (compile_flow, render_flow) or call render() directly in TypeScript.
Start, end, decision, process, subprocess, I/O, data, circle, note, manual, delay — each rendered with correct geometry.
Right-angle edges with rounded corners. Visio-style line jumps at crossings. Retry edges render dashed automatically.
Structured grid for top-down flows. Dagre for swimlanes and groups. Automatic fallback when the grid would explode.
Any OpenAI-compatible provider — OpenAI, Grok, Groq, Ollama. Bring your own key. No data leaves your infrastructure.
Assign nodes to actor lanes with #lane. Cross-lane edges route automatically between columns.
Clean, scalable output with CSS classes and data attributes — ready to embed, animate, or post-process.
compile_flow and render_flow tools over stdio. Drop FlowScript into any MCP-compatible AI agent.
Syntax highlighting, live SVG preview, zoom controls, light/dark theme. Runs entirely in the browser.
One Docker container. No cloud dependency, no vendor lock-in. Your data, your infrastructure, your LLM key.
Self-host the full stack — generator, editor, and LLM integration — with a single Docker container.
Spin up the full generator + editor stack. Bring your own LLM API key — any OpenAI-compatible endpoint.
services:
flowscript:
build:
context: https://github.com/
kilrkrow/flowscript.git
ports:
- "3000:3000"
environment:
LLM_API_KEY: your-key-here
LLM_BASE_URL: https://api.x.ai/v1
LLM_MODEL: grok-3-mini
Use FlowScript as a library. Import the renderer directly — no server required for pure SVG generation.
import { render } from 'flowscript';
const svg = render(`
#start Begin
Process Data
#decision Valid?
-> yes: #end Save
-> no: #end Reject
`);
// svg is a complete SVG string
Try the live editor — no account, no install, no catch.