AI Agent Memory across · shared project memory

AI Agent Memory across South Africa

AI coding agents forget everything the moment a session ends. MemHub turns each decision, fix, and constraint your team commits into durable project memory that every future session — human or agent — inherits automatically.

Every sprint produces knowledge that evaporates when the terminal closes: why an abstraction exists, which retry policy won, what production taught you last quarter. MemHub captures that context once and serves it back to every agent that touches the repository afterwards.

Create your shared memory

Free Community tier. Self-hosted or hosted. No external embedding API required.

acme/ai-agent-memory
synchronized

context/main

memories 41

pinned 6

contributors 9

BRIEFING FOR NEW SESSION

What changed since your last pull

3 new
01
decision

Standardize on idempotent write endpoints

Source: architecture.md

02
pattern

Repository layer owns every database transaction

Source: pull request #214

03
bugfix

Refresh tokens rotate atomically under load

Source: auth/session.ts

Claude Code
Cursor
Windsurf

LOCAL FIT

Why South Africa engineering teams choose MemHub

Across South Africa, software teams ship faster when their tools share one memory instead of three silos. New engineers ramp in days because precedent is searchable, not oral history.

Teams operating under POPIA can point at a concrete architecture instead of promises.

  • South Africa: connect Claude Code, Cursor, Windsurf, Cline, Codex CLI, or Antigravity to one shared memory. It also plugs into any CLI that speaks MCP or reads a JSON config.
  • Data residency you control — run MemHub on infrastructure inside South Africa or let us host it.
  • One command onboards the whole team: memhub login, then connect for Claude Code, Cursor, Windsurf, Cline, Codex CLI, or any JSON client.

UNIVERSAL COMPATIBILITY

Plugs into any CLI or development tool you use.

If it speaks MCP or reads a JSON config, MemHub plugs in. One command wires the majors; everything else joins as a generic client.

  • Claude Code
  • Cursor
  • Windsurf
  • Cline
  • Codex CLI (GPT)
  • GitHub Copilot
  • Google Antigravity
  • Gemini CLI
  • Aider
  • OpenCode
  • CI bots via REST

WHAT YOU GET

AI Agent Memory across South Africa, without the trade-offs.

01

MCP native

One stdio proxy exposes memory to any Model Context Protocol client without changing how your team works.

02

Searchable instantly

Full-text search out of the box, with optional embeddings through Ollama or a remote provider when you want semantic recall.

03

Persistent by default

Memories survive sessions, restarts, and device switches through a durable local outbox plus synchronized cloud storage.

04

Team-scoped

Organization and project boundaries are enforced at the API layer, so agents only see memory they are authorized to read.

05

Local-first writes

Work continues offline; entries commit to local storage first and synchronize when the network returns.

06

Provenance included

Every memory carries its author, source file or pull request, and version history — no anonymous tribal knowledge.

The honest comparison.

No feature-matrix gymnastics — just what changes the week you adopt MemHub.

Status quoWith MemHub
Context lives in heads and DMsMemory is structured, shared, and searchable
Every agent re-learns the repoSessions open pre-briefed with only what changed
Rules files cap out at ~2k tokensUnbounded versioned history stays one query away

Connect the tools already in your repository.

One command injects configuration for Claude Code, Cursor, Codex CLI, Google Antigravity, Windsurf, Cline, or any JSON MCP client.

repository terminal

$ npm install --global @simplexlat/memhub

$ memhub login

✓ project resolved · context ready · memory synchronized

Questions, answered straight.

What is AI agent memory?

It is a persistent store of decisions, constraints, fixes, and preferences that AI coding agents read at the start of every session instead of rediscovering — or hallucinating — them from scratch.

How is it different from a rules file like CLAUDE.md?

Rules files hold short-lived instructions inside one repo and one tool. A memory layer holds versioned, searchable, team-wide knowledge that travels across repositories, tools, and months of history.

What happens when our connection drops?

Nothing interrupts the work. Writes commit to the local SQLite outbox and synchronize automatically when the shared service is reachable again.

Can we try it without sending data anywhere?

Yes — the local engine works fully offline, and the self-hosted stack keeps even the shared layer on your own PostgreSQL.

How much does it cost?

A Community tier is free, and the Team plan adds multi-project contexts, seats, and higher synchronization limits. Self-hosting is unlimited.

MORE LOCATIONS

EXPLORE OTHER USE CASES

THE NEXT SESSION STARTS HERE

Stop rebuilding context.
Start building the product.

Create workspace