Potencia tu agente
Instalación del publicador. No es un endpoint de PlugMCP.
Host a copy with our keys and logs.
Aún no hay una versión empaquetada en el registro. Sigue el README del repositorio.
¿URL remota o un comando en tu máquina? En qué se diferencian las dos instalaciones
Ejecutar en PlugMCP
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
AbrirDel publicador
Your agents forget. Squish doesn't.
Your AI agent's memory lies to it. Every memory tool returns an answer — but most return the same confidence regardless of whether the answer is fresh, stale, or contradicted by a newer fact. The agent repeats plausible garbage as fact.
Squish gives AI agents calibrated recall confidence (ECE 0.055) and honest abstention — the only memory runtime that tells your agent how much to trust each recollection, and says I don't know when nothing clears the bar. Local-first, MIT, npm install in 30s.
Squish is the only memory runtime in this space that tells your agent how much each recollection deserves to be trusted, and says "I don't know" when nothing clears the bar.
npm install -g squish-memory && squish install --all. About 30 seconds, works on day one.One memory system for every agent. Claude Code -- Codex -- OpenCode -- MCP -- your own agents
npm install -g squish-memory && squish install --all
That is it. CLI, MCP server, and hooks for every coding agent on your machine. No API keys. No config. No Docker.
One package. Three interfaces.
npm install -g squish-memory
That's it. No separate SDK to install, no second package to version, no HTTP client to maintain.
If you need to access Squish from your own JS/TS code:
squish mcp --httpMCP is the standardized programmatic interface. There is no @squish/sdk.
Visual memory browser, knowledge graph explorer, and settings. Native installers for Mac, Windows, Linux.
Own this version forever. No subscription.
Your AI Tool (Claude Code, Cursor, etc.)
|
squish-mcp (MCP server)
|
squish-core (memory engine)
|
SQLite (.squish/squish.db)
squish remember "Use Drizzle ORM, not Prisma"
squish recall "What ORM should I use?"
squish search "database"
squish stats
Prepaid credit packs for sync and managed AI services. No subscription.
Credits buy: cloud storage, managed embeddings, LLM inference, remote sync.
Squish's HTTP server (squish mcp --http) binds to localhost by default and is intended for local use only. HTTP mode transmits data in plaintext -- do not expose it to the internet without a TLS-terminating reverse proxy (e.g., Nginx, Caddy, Cloudflare Tunnel).
For remote access:
All local data is stored in SQLite at .squish/squish.db with no encryption at rest. The database file permissions should be restricted on shared systems.
Offline, deterministic benchmarks on a 60-memory corpus. No LLM calls. Reproducible.
By query type:
Resurrection: 5/5 scenarios pass (dormant strengthening, irrelevant decay, accidental retrieval immunity, contradiction handling).
AGPLv3 -- Free to use, modify, and self-host. Commercial license available for proprietary embedding.
See LICENSING.md for details.
Squish gives AI agents calibrated recall confidence (ECE 0.055) and honest abstention — the only memory runtime that tells your agent how much to trust each recollection, and says I don't know when no
Resumen general
Squish gives AI agents calibrated recall confidence (ECE 0.055) and honest abstention — the only memory runtime that tells your agent how much to trust each recollection, and says I don't know when nothing clears the bar. Local-first, MIT, npm install in 30s.
Herramientas
Aún no hay lista de herramientas. Los servidores remotos la reciben tras la primera comprobación de estado.
Preguntas frecuentes
Copia en esta página la configuración stdio para Cursor, Claude Desktop, Claude Code o VS Code. Usa la instalación del publicador, no un proxy de PlugMCP.
Este registro no indica ninguna variable de entorno obligatoria.
Squish Memory se ejecuta localmente por stdio.
Última comprobación: 2026-11-08. Squish gives AI agents calibrated recall confidence (ECE 0.055) and honest abstention — the only memory runtime that tells your agent how much to trust each recollection, and says I don't know when nothing clears the bar. Local-first, MIT, npm install in 30s.