AgentNetwork-jev
Event-driven multi-agent network where Jev watches live state and assigns pending tasks to idle general-purpose agents.
Overview
JEV AgentNetwork explores a multi-agent topology in which Jev observes live network state and assigns pending work to available agents. The reference implementation, AgentRouter, lets Jev make routing decisions while general-purpose agents run work concurrently.
The network is event-driven rather than round-based: when an agent finishes, its completion event triggers a new routing decision right away instead of waiting for the team to synchronize. Agents are interchangeable by default and roles are optional prompt hints. The router sees pending work, recent results and which agents are idle or busy. The demo includes editable tasks, user-created workstreams, Todo steps and human-in-the-loop flows. It focuses on topology and scheduling, not on replacing an agent SDK.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which pending task should the idle agent take next?(inferred) | — |
How it works
- Agent completes a task and emits an event
- Router builds state: pending work, recent results, agent status
- Jev picks the next assignment
- Idle agent starts immediately while others keep working
Strengths
- No synchronization barrier between agents
- Fast enough to sit inside the network, per the listing
Caveats
- Reference implementation, not a full SDK
- Requires TypeSafe access
Getting started
Run the repository's start script, which installs the Python environment, starts AgentRouter and opens the demo.At a glance
- Type
- Project
- Status
- prototype
- Language
- Python
- Runtime
- local service + web UI
- Has code
- yes
- License
- Apache-2.0
Use cases and integrations
How curated lists describe it
JEV is fast enough to sit inside the multi-agent network—not just before it.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev decide?
Which pending task goes to which available agent, each time an agent finishes.
Are agent roles fixed?
No, agents are general-purpose; roles are optional prompt hints.
Is it production-ready?
It is a reference implementation focused on topology and scheduling.
Related projects
AutoGPT agent platform, listed as adding Jev decision blocks for typed routing, filtering, scoring and confidence-gated next actions.
ProjectproductionPythonOpenClaw plugin and CLI that turn a plain-language condition into a scheduled trigger, waking the model only when Jev says it is true.
PluginbetaTypeScriptProgrammable tool and agent runtime for Pi with one fabric_exec tool; includes an optional Jev loop for observe-decide-act.
ServicebetaTypeScriptLanguage and VM for long-running agent programs that pause for approval, resume after restarts and replay, with typed decision cells.
ProjectresearchTypeScriptOrchvia: local engine with TS and Python SDKs that runs Claude Code and Codex agents as a team, routing tasks by code and typed decisions.
LibrarybetaTypeScriptReusable project context and specialist workflows for existing AI tools, with an addon that offloads bounded decisions to Jev.
ProjectproductionTypeScript