Morion is a tool you and your agents need to
finish tasks bigger
than a session.
Notes and a task board on your computer that you and your AI agents share. Agents pick up tasks from the board and get the context they need. Morion can run the build, review and merge loop for you.
Keeping the first release small. Six pieces: data model, habit list screen, check-in flow, weekly review screen, local backup, app icon and onboarding.
Data model: diff ready, 3 files. Codex review passed!
Habit list screen: tests failed, weekly aggregation?! Re-opened with the failure attached.
Open on Habit Tracker:
- Habit list screen is in doing, second run after the weekly aggregation tests faileddecided
- Check-in flow is in todo, next in the queuedecided
- Weekly review screen, Local backup and App icon and onboarding wait in backlog, nothing scheduleddecided
Data model is done and merged to trunk.
Gap: Check-in flow mentions reminders, but no ticket covers notification permissions, and nothing in the notes decides local vs pushunknown
One agent is fine. Until the project gets bigger than a session.
With one agent and a simple project you open Claude Code, describe the change, review the diff, move on. The context fits in your head and in the window, and the CLI can even orchestrate a few subagents.
Then the project grows, and it breaks in several places at once.
- The codebase outgrows the context window.?!At some point the agent cannot read everything, so you feed it slices by hand: you paste prepared prompts and try not to forget the exceptions and the past experience. You become the retrieval layer, and you re-do it every session.
- Agents start losing context and tasks.?An epic has, say, twelve pieces with dependencies. Half is done in one session, then the agent compacts its context. You start getting bugs in things you discussed thirty minutes ago, and tasks that were planned quietly disappear.
- The work outgrows one agent.??You want Claude to write and Codex to review, or three tickets running in parallel. A moment later you are the dispatcher and the reviewer of every diff, in five terminals, with no record of who did what.
None of that is the agent's fault. It is good at a task. What it does not have is a place where the project lives: a board it can claim from, a long memory it can trust, and a loop that runs the build and the review without you in the middle.
That is the moment when Morion steps in.
What Morion adds to your workflow.
Organise notes and tasks, hand them to agents over MCP with the context they need, and run tasks through agents automatically.
Shared documents that your AI agents can read and edit together.
Like Apple Notes or Notepad, but with MCP that connects it to any CLI installed on your machine.
- Markdown, folders, tags, history and revisions
- Comments and advanced search
- Import and export .md files
Reminder architecture
Reminders are scheduled locally and mirrored to the server queue on sync. Decided: local first, server as backup. Push delivery stays out of scope until v2.
Open question: do queued deletes replay in order? See Queue writes while offline.
A kanban board where you and your AI agents create and manage tasks.
Assign a task to an agent in one click: it claims the ticket and works on it from start to delivery.
- Ticket claiming, so several agents never take the same one
- Task comments and resolutions, readable by agents and humans
- Replaces task management for a solo founder; sits below Jira or ClickUp in a team
Smart retrieval: every task with the context it needs.
When an agent picks up a task, Mo reads it, searches your notes and hands over what matters: related docs, tasks, lessons learned, restrictions, incidents, each with the reason it was picked. The same search answers your questions in Mo, the internal agent.
- Context picked at request time: no index to build or keep up to date
- Work packet: a task comes back with all its context in one call
- Secure: one AI access switch per folder keeps sensitive folders out of it
- Reads your notes only, not code, repos or anything else on your computer
Mo, the internal agent, in a simple chat.
Ask Mo about your projects and tasks and get an explicit answer with links to the notes it used. It can set the app up for you, build workflows, or create and review tickets on demand.
- A chat like Claude or ChatGPT, over your own notes and tasks
- Built-in tools to find information and set up the app
- Included in Pro, with no API key to set up
Yes, mostly. Reminders are scheduled locally and mirrored to the server queue on sync Reminder architecture decided. Push delivery is out of scope until v2 Reminder architecture decided.
Whether queued deletes replay in order is still open Queue writes while offline · comments provisional.
Conflict: the March spec says reminders are server-first Sync protocol v1 superseded; the decision note overrides it.
Gap: nothing in the notes covers reminder time zones. Want me to open a ticket for it?
Auto-code: a built-in code harness you start using in minutes.
Drag a task to Todo and it starts working. Your agents can manage and orchestrate the whole flow.
Build your own workflows: plain coding, the classic write-then-review with models from different providers, or advanced graphs with tens of agents, each with its own instructions.
- Workflows that run any task from the board
- Coding agents run on your Claude Code or Codex subscription, or on any model the agent supports
- Ask your CLI agent or Mo to build the workflow for you over MCP
- Spend caps per workspace
- Automated worktree-to-main merge, on demand
Local and MCP by design.
Runs on your machine, so any installed CLI connects to it securely, in parallel: Claude, Codex, Cursor and OpenCode work the same board at once.
- One-click setup per client
- Switch tool categories on and off: Read, Create, Update, Delete, Workflows
- Per-folder permissions: hide a folder from MCP entirely
| Actor | Calls | Last seen |
|---|---|---|
| mcp:claude-code | 1,204 | 2m ago |
| mcp:codex | 318 | 41m ago |
| mcp:cursor | 57 | Yesterday |
Pick the thing you want to do this week.
Each guide starts from the setup you already have and ends with something running. Ten to thirty minutes each.
Morion closes the gap between apps.
Agent boards run parallel sessions but keep no project memory. Memory SDKs store facts but have no board and no notebook. Notes apps with MCP hold your notes but do not pick the context a task needs or run a ticket. Harnesses run agents but have no place where the project lives. Morion puts the parts together in one local app.
| Morion | Notes with MCP | Agent boards | CLI coding agents | Memory SDKs | Agent harnesses | |
|---|---|---|---|---|---|---|
| Runs locally: all data on your machine | ||||||
| Markdown-native notes | — | — | ||||
| Persistent cross-session memory | — | |||||
| Hybrid / semantic search built-in | — | — | — | |||
| Work packets: each task with the notes it needs, picked at request time | — | — | — | |||
| Agentic Kanban board | — | — | — | |||
| Agent-claimable tasks (atomic) | — | — | — | — | — | |
| Internal AI assistant that manages data and performs actions (Mo) | — | — | — | — | ||
| Access for agents from different providers | — | |||||
| Folder and per-note AI permissions | — | — | — | — | ||
| Vendor-agnostic custom workflows (Auto-code) | — | — | — |