Install knowledge into your agent.
Each .mem is a typed knowledge graph: schema-validated on every write, sealed into one versioned file, installed in one command — and yours to keep as plain markdown in git.
●In active development. Memstead is pre-1.0 and built in the open — APIs, schemas, and the registry itself are still evolving. Expect a work in progress, not a finished product.
memstead install <scope>/<name> // find a mem below → install it → your agent queries it over MCP. New here? Get the CLI →
Mems
What is a .mem?
typed entities + relationships + a pinned schema
Not a notes export. Not an opaque index you cannot inspect. A .mem is a structured knowledge graph an agent installs and queries through MCP — every entity typed against a pinned schema that refused the writes which did not conform, and every entity's authoring rationale sealed inside the package, so you can ask why a claim is there without a running engine.
Is it safe to install a stranger's mem? A .mem is data your agent queries, not instructions it obeys: every entity carries an origin trust class, a foreign schema is served as structure only (its prose never enters your agent's instruction context), ingress is validated, and the mount is read-only. Treat third-party content as quoted data — the containment is structural, not a promise that no adversarial text can ever appear.
Share what you've built.
Package any mem as a .mem and publish it under your scope, authenticated with your GitHub identity. The registry is early — be one of the first publishers: a mem you ship is knowledge others install and query, not prose instructions they paste.
memstead publish --mem <name>