// documentation

These docs are generated from the memQL engine repository and versioned with each release — what you read here matches the code that shipped. They cover the harness, the language, the AI layer, operations, and the Cockpit.

Get started

Install the memQL Cockpit, connect to an engine, and run your first agent. This

Why memQL Is a Harness, Not a Library

Most "AI frameworks" hand you parts: a model client, a prompt

Get started

Get startedsite
Install the memQL Cockpit, connect to an engine, and run your first agent. This

Overview

memQL Quick Start
The app container npm installs two private SDK packages
memQL Tech Stack & Deployment Practices
This document establishes the opinionated technologies and practices for memQL development and deployment. These standards ensure consistency, effectiveness,…
memQL vs. Other Harnesses
This is meant to be fair. memQL is not the right tool for every job, and
Why memQL Is a Harness, Not a Library
Most "AI frameworks" hand you parts: a model client, a prompt

Concepts

MemQL Engine Architecture
MemQL is a domain-specific query language for time-series memory graphs, built on TimescaleDB. This document describes the modular engine architecture that…
Concept Seeding
Concept payloads can include optional seed.json files under each concept directory (for example, concepts/assistant/seed.json). These files describe the…
Concept ID Versioning
Concept IDs follow the format {version}:{domain}:{entity}[:{subtype}] where version
Data Validation Lifecycle
memQL uses a three-state validation lifecycle for data records: draft, checked, and confirmed. This replaces the old staging/production concept split with a…
MemQL Events System
This document describes the event pub/sub system in MemQL, which enables real-time notifications for graph mutations, queries, SI completions, and session…
Node Identifier Conventions
of memQL events / queries (including the CoPresent frontend).
Permissions and Access Control
This document describes the permission model and access control rules for conversations, assistants, attachments, groups, and system features.
Tool ↔ Knowledge Domain Pattern
extended 2026-05-08 to computer_use; further extended 2026-05-17

Language

MemQL Attribute Matrix
This document defines all @attribute decorators available in MemQL and their applicability to different definition types.
MemQL Authoring Rules & Gotchas
A running list of rules, conventions, and constraints that bite humans
MemQL Function Language Specification
MemQL functions provide reusable, parameterized operations. This specification defines the following function types:
MemQL
MemQL is the query and mutation language that powers the memory engine. It provides a deterministic, append-only interface for reading and writing…
MemQL Naming Conventions
Use receiver-specific prefixes so intent is obvious at call sites and in diffs.
MemQL Reserved Names
This document is the single index of every identifier MemQL reserves
MemQL Specifications
Specs are atomic, named boolean predicates declared in struct form:
Concept Cataloggen
Generated from the live DSL by cmd/docs-gen -- do not hand-edit.

AI

LLM cost control — defense in depth
This is the canonical map of how memQL bounds LLM spend so that a runaway
MemQL Operator Capabilities
This document is the single index of agent capability slugs and how

Operate

Access Model
memQL's authorization has three layers: authentication (who are
Actor envelope contract
When a request hits the gRPC stream, the auth interceptor
Identity Service (Operator Guide)
component/identity is the in-house authentication provider for
Node service-account JWTs
memQL cluster nodes authenticate to NodeService.Stream using an
Per-row authorization audit
memQL currently relies on partition-as-isolation-boundary for
Service-account JWTs (machine identity)
A service-account JWT is the machine identity for automation and synthetic
User Provisioning
How users get into the cluster. The identity service
Voice-agent service-account JWTs
The Go voice-agent (integrations/voice/agent/) authenticates to
Database Setup
memQL uses separate databases for each environment to ensure complete isolation and safety.
memQL Deployment Console -- Operator Guide
The Deployment Console is the admin/owner-only UI for driving the
memQL + CoPresent Deployment Strategy (AKS)
Authoritative deploy + operations reference for the memQL node mesh and the
Environment Variables -- memQL
memQL splits configuration into two tiers:
Infrastructure Management Guide
memQL + CoPresent run on Azure Kubernetes Service (cluster
Self-hosted LiveKit (staging/prod) — provisioning runbook
LiveKit powers CoPresent's realtime voice/video. We self-host the
Voice-agent bring-up verification
Operator runbook for the voice-agent's "is it actually working end-
Voice end-of-utterance (EOU) tuning
Status: baseline shipped; per-user adaptive endpointing is a design seed,
Workbench Runbook
Operational guide for the workbench capability -- the sandboxed
Workers — Operations Runbook
This document is the operator-side reference for the worker

Build

Audio Streaming Architecture
This document describes the audio paths in memQL: the audio
Build Tags -- Node Type Binaries
memQL uses Go build tags to compile separate binaries for each node type in the distributed cluster. Each binary includes only the code, integrations, and…

Everything memQL, from the source.

Get started

Overview

Concepts

Language

AI

Operate

Build