briqbriq
Dashboard

cookbook / memos

Memos

A markdown note store with an API, for memory a human can read and the agent can append to.

Run it

briq_run({
  recipe: "memos",
  params: {
    "size": "nano"
  },
  wait: true
})

Parameters

NameTypeOptionsDefault
sizeenumnano, small, medium"nano"

Outputs

Returned by briq_run once the stack is healthy. Hosts resolve inside your team network; secrets are generated per run.

MEMOS_URL
http://memos.<stack>.internal:5230

What it runs

BriqImagePortsSizeVolume
memosneosmemo/memos:stable5230nano5 GB at /var/opt/memos

When to use it

  • You want the agent's memory to be something you can read. A MEMORY.md grows until the agent has to re-read the whole thing to use any of it; Memos gives each fact its own note, with a timestamp, tags and a search endpoint.
  • The agent should append as it learns — one note per decision, per incident, per "this is why that test is flaky" — rather than rewriting one file each time.
  • You want to skim what your agent has been learning between sessions, from a browser, without asking it.

Quick capture is the whole design: short markdown notes, tagged, newest first. For structured recall instead of chronological, pair it with qdrant or neo4j. The volume is 5 GB at /var/opt/memos and holds SQLite plus uploads, so a stopped briq keeps everything.

Tips

  • Set up the first account through the UI. briq_expose port 5230 with mode: "token" and visit it once; the first account created becomes the host.
  • The agent writes through the REST API with an access token from Settings, one POST per memo. Tags are just #tag in the markdown body, so filtering by topic needs no schema.
  • Keep it nano. Notes are small; nano is the default and is enough for tens of thousands.
  • It pairs with the idle auto-stop. Memory that lives in a stopped briq costs nothing between sessions, and briq_start brings it back in about a second.