APLEXICA

A portable memory layer for every agent.

Install once; every agent you use reads and writes the same context, in each agent's own native format. No central server, no migration, no lock-in.

Get Started View on GitHub
Platforms
macOS · Linux · Win
Adapters
5 in V1

What lives in Aplexica.

  • 01

    Portable memories

    Memories live in a canonical, agent-agnostic format. Every supported agent reads them natively.

  • 02

    Conversation history

    Not just memories — your conversation history round-trips across agents. Pick up a thread in a different agent.

  • 03

    Skills and prompts

    Skill bundles you author once are available to every agent that supports them.

  • 04

    Real-time sync

    Multiple agents on the same machine see updates as they happen. No reload, no merge prompts.

  • 05

    Forking and branches

    Git-style branches across agents. Try the same prompt two ways. Continue with the answer you prefer.

  • 06

    Selective sync

    Rule-based filters decide which projects and memory types each agent sees.

Your files. Your agents. Our daemon stays out of the way.

Aplexica reads and writes the files each agent already uses. Memories live in your agent's native format on your disk. Uninstall Aplexica and every agent keeps its full state — there is nothing for us to hold hostage.

Aplexica filesystem-native architecture A three-layer stack: AI agents on top, the Aplexica canonical state in the middle, and each agent's native filesystem at the bottom. Coral dashed arrows show writes; cream solid arrows show reads. AGENTS · YOUR MACHINE Claude Code Codex Hermes APLEXICA · CANONICAL STATE ~/.aplexica agent-agnostic · versioned · forkable NATIVE FILES · EACH AGENT'S OWN FORMAT ~/.claude/ ~/.codex/ ~/.hermes/ writes reads
Fig 01 — Filesystem-native architecture

Three steps to multi-agent.

  1. 01

    Install

    A single command. The daemon detects the agents you already have and registers a per-agent adapter.

    curl -fsSL https://get.aplexica.com/install.sh | sh

  2. 02

    Use any agent

    Keep working in whichever agent you prefer. Aplexica watches files in the background and propagates changes.

    aplexica status

  3. 03

    Switch or fork

    Open a second agent and your full context is already there. Fork a conversation. Run agents in parallel.

    aplexica fork codex

Other tools unlock you by storing your state on their server — moving the lock-in, not removing it. Aplexica writes into the files each agent already reads.
Architecture brief aplexica/aplexica · docs/architecture.md

Built for trust.

  • 01

    Open source

    The daemon is auditable end to end. The whole thing is on GitHub.

  • 02

    Zero-knowledge cloud

    Cloud sync is end-to-end encrypted. Aplexica cannot read your content.

  • 03

    Local-first

    Default configuration makes zero network calls. Cloud is opt-in.

One install. Every agent.

Get Started Read the Docs