Privacy Policy
Archivum is local-first software: there are no accounts and no hosted storage. This page covers two things: the Archivum command-line tool and MCP server that you run on your own machine, and the Archivum website, including its simulation. “We” means the operator of this website. Contact: hello@archivum.dev.
The Archivum CLI and MCP server
They store everything as files on your computer, in ~/.archivum or the folder in $ARCHIVUM_HOME:
- Notes that your agent or you save, and transcripts of agent sessions, written by the Claude Code hooks or
archivum log. Known secret patterns such as API keys and tokens are replaced with a[REDACTED]placeholder before they are written, but this cannot catch everything. - The wiki of markdown pages, the history of every write to it, and the proposals that dreams produce while you review them.
We do not receive any of this. The software has no analytics, telemetry or tracking, and does not contact us. It makes network requests only when you run a dream.
Dreaming and your LLM provider
When you run archivum dream, the transcripts, notes and wiki pages that the dream needs are sent to the LLM endpoint you configured: Anthropic by default, or any OpenAI-compatible endpoint you point it at, which may be a local router on your own machine. That provider receives your data under your own agreement and API key, and what it does with it is governed by its own terms and privacy policy, for example Anthropic’s privacy policy. If you configure no key, nothing is sent anywhere. Do not dream over text you are not allowed to share with your provider.
The website simulation
The page at /try/ simulates memory and dreaming in your browser. Its recorded mode uses a scripted example and sends nothing. In live mode, the text you type in the chat, and the memory text the simulation builds from it, is sent to this website’s server, which forwards it to a model provider and returns the reply. The server stores none of it: it keeps no database, accounts or chat logs, and its logs do not contain request bodies. For rate limits and daily usage caps it keeps counters per IP address in memory, which are reset daily. The model provider that handles the request has its own policies, so do not type anything private or secret into the simulation.
The website generally
We do not use analytics, advertising or tracking scripts, and the site sets no cookies. Whoever hosts the website, and any network in front of it, may keep ordinary request logs such as IP address and time.
Deleting your data
- Everything the CLI stores is in its folder. Delete files there, or the whole folder, at any time.
archivum retentiondeletes transcripts that were dreamed more than a given number of days ago (30 by default); dream and apply also run it with the default. Transcripts that no dream has processed stay until you delete them.- Data already sent to your LLM provider is held under that provider’s terms; ask them to delete it.
Your rights
Because we hold no data about you, there is nothing for us to export or correct. For any question, write to hello@archivum.dev.
Children
Archivum is not meant for anyone under 16.
Changes
If we change this policy in a way that matters, we will change the date above.