Why we built this

Tycana exists because we believe something specific: text is the universal interface, and the tools that understand this will outlast the ones that don’t.

That sounds abstract. Here’s what it means in practice.

The problem with productivity tools

Every task manager on the market was built for the same interaction model: open a browser, click through a GUI, drag things around. Some of them are genuinely beautiful. Todoist, Things 3, Linear — real craft went into those products.

But they share an assumption that’s becoming a liability: the human is always the one at the keyboard.

When you type a task into Todoist, you’re filling out a form. When your AI assistant tries to add a task to Todoist, it has to navigate an API that was designed as an afterthought to the GUI. The tool’s architecture — built for mouse clicks and form fields — fights programmatic interaction at every level.

So these tools bolt on chatbots. Add copilot sidebars. Build wrapper APIs. They’re trying to teach a GUI to speak text.

We went the other way.

Text-native by design

Tycana was built for text from the beginning. Not because we’re nostalgic for the terminal — because we saw where things were heading.

When you type tycana add "Review PR tomorrow 2pm", you’re expressing intent through structured text. When an AI assistant says “add a task to review the PR by tomorrow afternoon,” it’s expressing the same intent through structured text. Same language. Same semantics. Different speaker.

This isn’t a coincidence. It’s the whole point.

A well-designed CLI tool is inherently AI-ready. Structured commands. Predictable output. Clean semantics. These are the same properties that make a tool work well with the Model Context Protocol — the standard that Claude, GitHub Copilot, and the next generation of AI assistants use to interact with external tools.

When MCP emerged, our integration was almost trivial. We didn’t have to build an adapter layer between our real interface and an AI-friendly one. They were the same interface.

We didn’t add AI to a task manager. We built a tool that naturally speaks the language AI understands.

The terminal renaissance

The terminal is having a moment, and it’s not nostalgia. It’s because AI arrived and AI lives in the terminal.

Claude Code. Codex CLI. Gemini CLI. The most powerful AI coding tools of 2025 all launched as terminal applications. They all support MCP. They all want to interact with your tools programmatically.

Developers are spending more time in their terminals than they have in years — not because the GUI failed, but because the terminal is where the AI agents are. And those agents need tools they can talk to.

The MCP ecosystem grew from roughly 100 servers in late 2024 to over 8,000 by early 2026. It’s backed by Anthropic, OpenAI, Google, Microsoft, and the Linux Foundation. This isn’t a trend. It’s infrastructure.

Tycana sits at the exact intersection: a CLI tool that’s also an AI agent interface. Not a web app with a CLI bolted on. Not an API wrapper pretending to be a tool. A genuine command-line application that happens to be exactly what AI assistants need.

Three interfaces, one reality

Most tools think in terms of features. We think in terms of interfaces.

Your tasks live in one place. You access them from whichever interface fits the moment:

Your terminal — where you work. Add tasks with natural language. Filter by project, tag, or date. Script it, automate it, pipe it. Sub-100ms. No GUI to load.

Your calendar — where you plan. Tasks with times become calendar events automatically. Standard ICS feeds that work with Apple Calendar, Google Calendar, Outlook — anything that reads an .ics URL. No integration to configure. No OAuth flow.

Your AI assistant — where you think. Ask Claude what’s overdue. Tell it to add a task. Say “mark the deploy task done.” Your actual AI assistant, the one you already use for everything else, now has full access to your tasks.

These aren’t three products stitched together. They’re three windows into the same reality. Add a task in your terminal at 11pm, see it on your calendar the next morning, reschedule it through Claude over coffee. Same task. No sync to think about. No workflow to maintain.

What we believe

The architecture is the product. It’s not about any single feature — it’s about building something where terminal, calendar, and AI are natural consequences of the design, not bolted-on integrations.

Calm tools win. People abandon task managers because managing the system becomes work. Tycana should be fast, invisible, and forgettable until you need it. Add a task, move on. The system handles the rest.

Your data is yours. Tasks are plain YAML files on your machine. No database. No proprietary format. If Tycana disappeared tomorrow, your tasks would still be readable text files. That’s not a limitation — it’s a promise.

Free means free. The CLI is complete without an account. Unlimited tasks. Natural language. Recurring tasks. Local YAML storage. When you want the connected experience — sync across devices, calendar subscriptions, AI access — that’s the Sync tier. $6/month. No trial. No features held hostage.

Simple beats impressive. We’d rather have a tool that does fewer things well than one that does everything poorly. Every feature earns its place or gets removed.

Try it

brew install tycana/tap/tycana

You’ll add your first task in under a minute.