Made by 100monkeys AI, Inc.
Write once. Read everywhere — from your editor, from your team, from every LLM you give access to. A persistent, semantically navigable knowledge substrate that outlives the model behind it.
Let your AI sign you up
Nuclear Notes ships an Anthropic Agent Skills manifest at the URL below. Any skill-compatible client will pick a subdomain, register your instance, and send you a claim link.
https://cortex.page/skillRead https://cortex.page/skill and follow the instructions to create a Nuclear Notes instance for me.You'll verify ownership via magic-link email before the instance is yours.
For people who outgrew their tools
Personal notes
The same cortex your LLM reads. No plugins, no copy-paste, no vault lock-in. Cross-links, semantic neighbors, and revision history come with the substrate.
Replaces: Obsidian, Roam, Logseq, Apple Notes, Notion
Research
Drop in a paper or a book. Traverse its conceptual neighborhood without re-reading the shelf. Surface insights the author didn't intend by following the embedding graph.
Replaces: Zotero, Mendeley, paper folders, scattered PDFs
Team wikis & code docs
Real-time updates as the codebase changes. Engineering agents write the diff; humans review the history. Code documentation that doesn't rot.
Replaces: Notion, Confluence, GitHub Wiki, README files
LLM memory
Persistent. Auditable. Revocable. Portable across every LLM client via Model Context Protocol — one token, every model, same knowledge.
Replaces: ChatGPT memory, Claude Projects, homegrown RAG
How it works
Pages, atoms, and structured metadata. The same rich editor humans and agents read from — one substrate, no separate document store.
Issue an MCP token from settings. Claude Desktop, Cursor, Continue, your own agent — anything that speaks Model Context Protocol attaches in one paste.
Semantic neighbors surface the next idea. Agents update structure and metadata as your understanding grows; every action is permission-checked, audited, reversible.
Capabilities
Notes, links, embeddings, and revision history live in your substrate, not in a vendor's weights. Swap the LLM behind the screen; the cortex is still yours.
Embeddings on every page and atom. Walk from one idea to its conceptual neighbors without rereading dense material — research libraries, books, codebases all navigate the same way.
Reusable fragments transclude across pages and update everywhere they appear. The structure mirrors how thinking actually composes — small definitional nuclei stitched into long-form arguments.
Seventy-plus MCP tools mounted at the instance, no integration code required. Every action permission-checked, audited, reversible. Agents are first-class citizens, not bolt-ons.
Attach typed key-value data alongside any page or atom — status, owners, sources, custom schemas. Filterable from the editor, queryable from MCP. Curate the cortex the way you actually think.
Wire an agent to your repo and the docs update as the code does. Architecture decisions, runbooks, API references — kept current by the agent that watches the diff, reviewed by the human that opens the PR.
Pricing
$0forever
Starting at
$12/ mo.
Billed monthly
Starting at
$36/ mo.
Billed monthly
Custom pricing
One token. Every model. Audited and revocable. Get started in under five minutes — the cortex is the product surface.