molecule-ai/molecule-core
35
0
Fork 2
Code Issues 16 Pull Requests 2 Actions 49 Packages Projects Releases Wiki Activity
409a249ca6
molecule-core/docs/product/overview.md
Hongming Wang 24fec62d7f initial commit — Molecule AI platform 
Forked clean from public hackathon repo (Starfire-AgentTeam, BSL 1.1)
with full rebrand to Molecule AI under github.com/Molecule-AI/molecule-monorepo.

Brand: Starfire → Molecule AI.
Slug: starfire / agent-molecule → molecule.
Env vars: STARFIRE_* → MOLECULE_*.
Go module: github.com/agent-molecule/platform → github.com/Molecule-AI/molecule-monorepo/platform.
Python packages: starfire_plugin → molecule_plugin, starfire_agent → molecule_agent.
DB: agentmolecule → molecule.

History truncated; see public repo for prior commits and contributor
attribution. Verified green: go test -race ./... (platform), pytest
(workspace-template 1129 + sdk 132), vitest (canvas 352), build (mcp).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-13 11:55:37 -07:00

2.8 KiB
Raw Blame History

Overview

What Molecule AI Is

Molecule AI is an org-native orchestration platform for AI agent workspaces.

The shortest accurate description is:

A visual org chart plus a control plane for heterogeneous agent teams.

Instead of modeling a system as edges between tasks, Molecule AI models it as roles inside a hierarchy. A workspace can be one agent now, a sub-team later, and still keep the same external identity, policy boundary, memory boundary, and position on the canvas.

What Problem It Solves

Most agent projects are strong at one of these layers, but weak across all of them together:

  • runtime flexibility
  • topology management
  • memory isolation
  • operational control
  • observability
  • reusable skill lifecycle

Molecule AI is the layer that ties those together.

What Makes It Different

Dimension Typical agent tool Molecule AI
Primary abstraction task, chain, graph node workspace role
Topology manual edges or hard-coded routing org chart hierarchy
Runtime choice usually one framework multiple frameworks behind one workspace contract
Memory model flat or loosely namespaced hierarchy-aware scope + awareness namespace
Team growth rebuild the graph expand a workspace into a sub-team
Ops mostly left to custom glue built-in registry, heartbeats, traces, approvals, activity, restart

Runtime Compatibility

Current main ships adapters for:

  • LangGraph
  • DeepAgents
  • Claude Code
  • CrewAI
  • AutoGen
  • OpenClaw

Branch-level runtime work such as NemoClaw exists separately and should be described as WIP, not merged main support.

Memory And Skills

Molecule AI treats durable memory and reusable procedure as different system layers:

  • memory stores facts worth recalling later
  • session-search recovers recent activity and memory rows
  • skills store repeatable procedures
  • promotion is the bridge: repeated durable workflows can be elevated from memory into a hot-reloadable skill package

This separation is one of the reasons Molecule AI scales better than “just add another memory store” designs.

What Molecule AI Is Not

  • Not a replacement for LangGraph, CrewAI, AutoGen, Claude Code, or OpenClaw
  • Not a visual workflow automation builder where nodes are one-off tasks
  • Not just a chat UI over one agent
  • Not a model provider
  • Not a hosted SaaS-only black box; this repository is the open-source core

Related Docs