Back to directory

Synapse

A self-hosted AI workspace where conversations are shared team rooms for people, native actors, and bridged remote agents, with governed access to plugins, MCP tools, memory, and local devices.

Visit website (opens in a new tab)synapse-zai-org.vercel.app
Synapse

Synapse is a conversation-centric runtime for digital teammates. Rather than treating chat as an interface to an isolated bot, it treats each conversation as the collaboration boundary: humans, platform-native actors, and bridged remote agents share one thread, and everything they can touch, including plugins, skills, devices, sandboxes, event sources, and memory, is governed at the workspace layer through explicit, revocable grants.

Four kinds of participants can share a thread: workspace members, native actors, bridged remote agents, and external IM identities. Actors can wake each other so multi-agent handoffs happen inside the visible transcript. A machine-side daemon bridges existing coding agents such as Claude Code and Codex CLI into conversations over an outbound WebSocket, pulling messages and posting replies through a per-conversation tool surface rather than receiving pushed content.

Synapse is written in TypeScript and licensed under Apache-2.0. It is described as an early design, with schemas and runtime contracts still subject to change.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.