Hi HN, I’m Scott. I built marktoflow because I reached my limit with the "node-spaghetti" and proprietary UIs of drag-and-drop automation tools.
marktoflow is a CLI-native framework that defines automation logic using Markdown + YAML. The goal is to move the "thinking" part of automation out of opaque web builders and into your local development environment.
Why this approach?
1. Markdown + YAML (Diffable & Version Controlled): Decisions are defined in plain text. This makes your workflows fully version-controllable, diffable, and incredibly easy for LLMs to generate. No more opaque "black box" logic hidden in a database.
2. CLI Native / BYO Interface: You don't need a heavy browser-based builder. You can build and manage your flows directly using GitHub Copilot, Claude Code, or Codex. It fits into the tools you already use every day.
3. GUI Visualization: While it's CLI-first, it includes a GUI specifically for visualizing the flow logic, making it easy to audit complex decision paths at a glance.
4. AI-Native Decisions: Most tools are "pipes" (A to B). marktoflow is built for "decisions"—triaging data, drafting content, or analyzing context based on LLM rules defined in your Markdown.
I’ve found that the best "builder" for marktoflow isn't a website—it's Claude. I just describe the business logic I need, and Claude builds the entire Markdown/YAML workflow file for me.
It’s in early alpha and I'm currently using it to automate the triage and response drafting for my side projects. I'd love feedback on the framework structure and the text-first approach to AI orchestration!
Hi HN, I’m Scott. I built marktoflow because I reached my limit with the "node-spaghetti" and proprietary UIs of drag-and-drop automation tools.
marktoflow is a CLI-native framework that defines automation logic using Markdown + YAML. The goal is to move the "thinking" part of automation out of opaque web builders and into your local development environment.
Why this approach?
1. Markdown + YAML (Diffable & Version Controlled): Decisions are defined in plain text. This makes your workflows fully version-controllable, diffable, and incredibly easy for LLMs to generate. No more opaque "black box" logic hidden in a database. 2. CLI Native / BYO Interface: You don't need a heavy browser-based builder. You can build and manage your flows directly using GitHub Copilot, Claude Code, or Codex. It fits into the tools you already use every day. 3. GUI Visualization: While it's CLI-first, it includes a GUI specifically for visualizing the flow logic, making it easy to audit complex decision paths at a glance. 4. AI-Native Decisions: Most tools are "pipes" (A to B). marktoflow is built for "decisions"—triaging data, drafting content, or analyzing context based on LLM rules defined in your Markdown. I’ve found that the best "builder" for marktoflow isn't a website—it's Claude. I just describe the business logic I need, and Claude builds the entire Markdown/YAML workflow file for me.
It’s in early alpha and I'm currently using it to automate the triage and response drafting for my side projects. I'd love feedback on the framework structure and the text-first approach to AI orchestration!