How do I manage my local Wiki?

Browse and edit your Markdown Wiki, organize projects, topics, people, prompts, and tasks, follow backlinks, and trace notes to their source sessions.

The Local Wiki is a local Markdown knowledge base for projects, topics, people, prompts, and tasks. When AI extraction is enabled, OneMira can suggest updates or write them according to the selected auto-index mode.

Where do I browse the Local Wiki?

Navigate your knowledge graph with the sidebar, and read or edit pages in the distraction-free detail view.

PEOPLE · 21
PROJECTS · 1
Feature Spec: Canvas...
2h
PROMPTS · 4
REFERENCES · 10
TASKS · 1
TOPICS · 23
Feature Spec: Intelligent Canvas UI

Voice note captured during the morning walk. The traditional chat UI is too linear. We need to upgrade to a spatial, Notion-style block canvas where users can visually arrange AI-generated components.

Instead of outputting plain markdown, the LLM must stream structured JSON. The frontend will parse this stream and render interactive React components dynamically on the grid.

Architecture Decisions
  • State Management: Rip out Redux. Move to Zustand for the canvas nodes to prevent React re-render thrashing during high-frequency WebSocket streaming.
  • Animations: Use Framer Motion for layout transitions. Block insertions must hit 60fps (sub-16ms frames) to feel buttery smooth.
  • Drag & Drop: Integrate @dnd-kit/core
Cursor Execution Context
  • Task 1: Scaffold a new `CanvasStore` in Zustand with `addBlock`, `updateBlock`, and `moveBlock` reducers.
  • Task 2: Build the `BlockRenderer` component that dynamically imports UI elements based on the incoming block's `type` field.

How do I maintain Wiki content?

Use explicit agent instructions

External AI clients may use files such as CLAUDE.md or AGENTS.md for project-specific instructions. This behavior belongs to the selected client; OneMira does not create or enforce those files automatically.

# Project Context Synchronization
## Auto-Sync Action Rules

Whenever a new voice note is synced to /raw:
- Extract Action Items into tasks.md
- Cross-link pages with [[Wiki Links]]

Run a verified extraction job

After C1 sync and local transcription finish, the session is available as Markdown. You can start a configured AI extraction job from the app or CLI and then review its proposed Wiki changes.

$ mira session list --limit 5
$ mira ai run <prompt-id> --session <session-id>
$ mira wiki list --kind topics

Compounding Queries

A connected AI client can compare source sessions and draft a durable Wiki note. Ask it to show the proposed Markdown and source session IDs before writing the file.

USER: Compare the auth sessions and cite each session ID.
AGENT: I found two relevant sessions. Here is a sourced draft.
USER: Write the reviewed draft to wiki/topics/auth-comparison.md.

Readable files with a local index

Wiki content is stored as plain .md files on your computer. OneMira also maintains a local SQLite-backed index for search and application state. You can inspect the Markdown in Obsidian or VS Code and version it with Git.

ObsidianObsidian
VS CodeVS Code
GitGit