Notes
Short memories the agent saves itself.
remember update_note forget list_notes
Nexorum keeps what your agent learns as plain files on your machine: short notes, session transcripts and a small linked wiki. A background dream reviews them, corrects stale facts and proposes better memory. You read the diff and approve it.
$ npx nexorum initPosted here at launchFree and open source. No account. Nothing leaves your machine until you dream.
17 MCP tools and a context resource. Notes are searchable right away; dreams turn them into a wiki you review.
Short memories the agent saves itself.
remember update_note forget list_notes
BM25 search over notes and wiki pages.
recall search_documents
A short digest of the wiki, loaded at session start.
context nexorum://context
Plain markdown pages, linked from an index.
read_document list_documents related_documents document_history
Transcripts for later dreams, and the state of memory.
save_transcript memory_status
Review proposals, or dream through your agent with no key.
list_proposals show_proposal dream_prepare dream_submit
Nexorum is open source software that runs on your machine. There is nothing to sign up for. Notes, transcripts, the wiki and its history are files under ~/.nexorum; delete the folder and the memory is gone.
The only thing that costs money is the model you choose to dream with, and you pay that provider directly. A dream sends transcripts, notes and wiki pages to the LLM you configure, and nowhere else.
| The CLI and MCP server | MIT licence, runs on your machine | Free |
| Dreaming | your key, your budget | Your provider |
| Accounts, servers, telemetry | there are none | None |
nexorum init sets up the hooks and MCP server for the clients it finds. Or wire things by hand, as shown here. Select a card for what it changes.
# one command, every client it finds: # Claude Code, Codex, Hermes, # Cursor, Claude Desktop npx nexorum init # optional: install the command npm install -g nexorum # later: review what your # sessions taught the agent nexorum dream nexorum apply
# the MCP server only claude mcp add nexorum -- \ npx -y nexorum mcp # plus the hooks npm install -g nexorum nexorum init --claude-code \ --prompt-recall
codex mcp add nexorum -- \ npx -y nexorum mcp # so Codex actually uses it: # a short block in this # project's AGENTS.md nexorum init --codex
# init does this for you # when it finds Hermes; by hand: hermes mcp add nexorum \ --command npx \ --args -y nexorum mcp
{ "mcpServers": {
"nexorum": {
"command": "npx",
"args": ["-y", "nexorum", "mcp"]
}
} }Your agent writes notes and its sessions are saved as transcripts. A dream looks back over them with hindsight and proposes updates to a small wiki. The next session starts from the approved version.
Claude Code hooks or the MCP server save short notes and session transcripts to ~/.nexorum. Known secret patterns are redacted before anything is written.
Notes and wiki pages are ranked with BM25 keyword search, and the agent can follow links between pages from the index. No embeddings, no vector database.
nexorum dream checks the wiki against your transcripts, fixes stale or wrong facts, merges duplicates and links pages. It writes a proposal; nothing changes until you apply it.
A short index of the wiki is loaded at session start, so your agent does not have to rediscover what earlier sessions already learned.
$ nexorum remember "Run tests with pnpm test:unit, not npm test" Saved note note_7Qm2xK9vTq. $ nexorum dream Pending transcripts: 3 Unconsolidated notes: 4 Rough estimate: ~180,000 tokens (budget 3,000,000). This is billed by your LLM provider. Proceed? [y/N] y Proposal prop_8f2c41 (review): 2 added, 2 modified, 0 removed added /ci.md added /config.md modified /testing.md modified /_index.md $ nexorum diff ~ /testing.md - Tests: `npm test` (jest) + Tests: `pnpm test:unit` (vitest), since 2026-11-19 $ nexorum apply Applied: 4 path(s) written.
nexorum apply.Notes are searchable right away. Dreams turn notes and transcripts into a small wiki and propose fixes. You decide what is applied, and everything stays in files you own.
Plain markdown files in ~/.nexorum, linked to each other, with an _index.md that lists them. Open them in any editor, or let the agent follow the links.
Each write to the live wiki, from a dream you applied or from you, is recorded. See the history of a page and restore an earlier version.
A dream only writes a proposal. Nothing reaches the live wiki until you run nexorum apply, and a page changed since the dream started is skipped.
Use your own key (Anthropic by default, or any OpenAI-compatible endpoint), or let the agent you already run dream through two MCP tools. Key runs are estimated first and stop at your budget.
API keys, tokens and private keys are replaced with [REDACTED] markers in transcripts, notes and anything a dream writes, before they reach disk.
Read the line diff against your live wiki, then apply or reject it. The agent can read proposals but never apply them.
~/.nexorum, one space per project, as plain files. Open them in any editor, commit them to git, or delete the folder.Or watch it work first: the simulation takes about a minute and runs in your browser.
$ npx nexorum initPosted here at launchLocal: notes, transcripts, wiki and history are files under ~/.nexorum. Dreaming sends them to the LLM you configure, and nowhere else.