Skip to content
Atomic

Atoms

Atoms are the fundamental unit of your Atomic knowledge base. Each atom is a markdown note that can be searched, tagged, linked, cited, and visualized.

What Is an Atom?

Section titled “What Is an Atom?”

An atom contains:

  • Content - Markdown text.
  • Title - Derived from the content or set manually.
  • Snippet - A short preview for lists and search results.
  • Source URL - Optional link to the original source.
  • Source - Capture origin such as manual, imported, ingested URL, feed, or integration-created content.
  • Published date - Optional timestamp for articles and imported material.
  • Tags - Hierarchical labels, auto-extracted or manually assigned.
  • Pipeline status - Embedding and tagging state for background processing.

The Processing Pipeline

Section titled “The Processing Pipeline”

When you create or update an atom, Atomic can run an asynchronous pipeline:

  1. Chunking - The content is split at markdown-aware boundaries such as headers, paragraphs, and code blocks.
  2. Embedding - Each chunk is sent to the configured embedding provider.
  3. Tagging - The full content is analyzed by an LLM to extract structured tags when auto-tagging is enabled.
  4. Edge building - Vector similarity is computed against other atoms to find semantic relationships.

This pipeline is fire-and-forget from the caller’s perspective. You get the saved atom immediately while processing continues in the background.

Each atom tracks embedding_status and tagging_status. Common values are pending, processing, complete, and failed; tagging can also be skipped when auto-tagging is disabled or not applicable.

Creating Atoms

Section titled “Creating Atoms”

You can create atoms in several ways:

  • Editor - Write directly in Atomic’s markdown editor.
  • Web Clipper - Use the Atomic Web Clipper to save web pages or selected text.
  • API - POST /api/atoms with markdown content.
  • Import - Bulk import an Obsidian vault.
  • URL ingestion - Ask Atomic to fetch and extract an article from a URL.
  • RSS/Atom feeds - Subscribe to feeds and let Atomic create atoms for new entries.
  • MCP - Let an MCP client create or update memory using Atomic tools.
Section titled “Markdown Links”

Atoms can contain wiki-style links using [[...]]. Atomic stores discovered links separately so the editor and graph views can resolve relationships between atoms while preserving unresolved targets for future editing.

Duplicate Source URLs

Section titled “Duplicate Source URLs”

URL ingestion and bulk capture paths can skip content when an atom with the same source URL already exists. This prevents feed polling and browser clipping from creating repeated copies of the same article.

Section titled “Related”