Pinned Tweet
Archivum officially live on @ponsdotfamily.
CA: 0xda4283f14b487034b90d865aac489994f2c680c4
Give your AI agent a memory it keeps.
Archivum keeps what your agent learns as plain files on your machine: short notes, session transcripts and a small wiki. A background “dream” reviews them, corrects stale facts and proposes better memory. You read the diff and approve it.
tryarchivum.com
Review, diff, apply or reject, A dream only writes a proposal. Read the diff against your live wiki, then apply or reject it. A page changed since the dream started is skipped, not overwritten.
4
142
Dream with a key, or without one, Use your own key (Anthropic by default, or any OpenAI-compatible endpoint, including a local router), or let the agent you already run do it through two MCP tools, so no separate key is needed. Either way the result is a proposal you apply yourself. Key runs are estimated first and stop at the token budget you set.
2
5
156
I understand the concern, but that's not what we're aiming for.
We do plan to build real utility for ARCH. The current version of Archivum is free and open source, and we're focused on making the product genuinely useful first.
We're working toward connecting ARCH to actual ecosystem activity, with utility that gives people a reason to use the token beyond speculation.
We haven't finalized the mechanics yet, and we'd rather take the time to build something meaningful than announce utility just for the sake of it.
Judge us by what we build and deliver. We're committed to giving ARCH a real purpose.
Every change is a version. 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.
every time you start a new coding session, your AI agent forgets what you worked on yesterday.
the decisions, the fixes, the details you already explained. Gone.
We built Archivum to change that. It gives your agent memory as plain Markdown files on your machine. No account, no server, no database.
tell it a fact once and it carries across sessions. When a fact changes, the old note is updated instead of stacking contradictions. Known secret formats are scrubbed before anything is saved.
then there's the dream: it reads your notes and sessions and proposes pages for a project wiki. Nothing is applied until you read the diff and approve it.
i recorded the whole thing in claude code with its built-in memory turned off.
open source : github.com/tryarchivum/archi…
npx archivums init
tryarchivum.com
full demo below ↓
Good question, and you're right to point that out.
Today, ARCH is not required to use Archivum. The CLI, MCP server, and memory features are free and open source. Developers don't need to buy, hold, or spend ARCH to use them.
We don't want to introduce artificial token requirements that make the product harder to adopt.
Token utility is something we're still working on. We haven't finalized which feature will use ARCH first, so I don't want to promise a specific mechanism before it's ready.
Our direction is to make any future utility connected to actual ecosystem activity, rather than forcing developers to hold a token just to access basic memory features.
As Archivum adoption grows, we want to explore ways for that activity to create meaningful demand for ARCH. But that needs to come from real utility, not speculation.
For now, our priority is making Archivum useful, growing adoption, and building the ecosystem first.
Replying to @Use_Archivum
What can I use ARCH for in Archivum today? I tested the memory tools without needing ARCH. Does any feature require developers to buy, hold, or spend it? If that’s still on the roadmap, which feature will use ARCH first, and how would wider adoption create demand for the token?
A wiki your agent can navigate. Plain markdown files in ~/.archivum, linked to each other, with an _index.md that lists them. Open them in any editor, or let your agent follow the links and search.
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.
Replying to @Use_Archivum
I tested Archivum and got the same note back through MCP after restarting. Pretty cool. @Use_Archivum, could you share a tiny demo repo that lets any dev connect their own agent, teach it something, then start a fresh session and ask what it remembers? I’d love to see what other devs try with it.
Install Archivum in one command!
Run archivum init and it sets up the hooks and MCP server for the clients it finds. You can also wire things by hand, as shown here.
- Hooks verified against the Claude Code docs
- Standard MCP over stdio, 17 tools
- One space per project, kept as plain folders
- Not yet tested in every MCP client
Read the quickstart: tryarchivum.com/docs/#quicks…
From a conversation to memory that gets better:
1. Your agent works
Claude Code hooks or the MCP server save short notes and session transcripts to ~/.archivum on your machine. Known secret patterns are redacted before anything is written.
2. Notes are searchable
Notes and wiki pages are ranked with BM25 keyword search. The agent can also follow links between wiki pages from the index. No embeddings, no vector database.
3. A dream reviews and corrects
When you run archivum dream, a model checks the wiki against your transcripts, fixes stale or wrong facts, merges duplicates and links related pages. It writes a proposal, and nothing changes until you apply it.
4. The next session starts informed
A short index of the wiki is loaded at session start, so your agent does not have to rediscover what earlier sessions already learned.
tryarchivum.com
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.
AI agents can be incredibly smart within a session, but once the session ends, they become amnesiac.
Archivum solves this by giving AI agents long-term memory stored as plain files directly on your machine.
While working, agents save short notes as memories.
But not every note is necessarily correct.
For example, an agent might guess a configuration file path without ever verifying it.
Later, you tell it that the project has moved to pnpm and Vitest.
Now, the old notes and new information contradict each other.
This is where **Dream** comes in.
Dream revisits conversation transcripts and existing notes with better context.
It then:
- Verifies information against your wiki.
- Corrects outdated facts.
- Merges duplicate notes.
- Resolves conflicting information.
The interesting part? Dream never modifies your memory directly.
Instead, it proposes changes. You review the diff and decide whether to approve or reject them.
What Archivum offers:
- 17 MCP tools and 18 CLI commands.
- Secrets are redacted before anything is written to disk.
- Every wiki change is versioned and can be restored.
- Run Dream with your own LLM API key or through an agent you already use, with no additional API key required.
Archivum is open-source under the MIT license, with zero runtime dependencies.
Get started:
`npx archivums init`
Or try the simulation without installing anything:
tryarchivum.com/try




