Viktor.comAn AI coworker that actually does the work
Promoted
h5i-python
Python SDK for Programmable Multi-Agent Orchestration
6 followers
Python SDK for Programmable Multi-Agent Orchestration
6 followers
Multi-agent coding is more than launching agents or letting them chat. h5i-python lets you program the full development process in Python: Claude Code, Codex, and other runtimes can implement independently, review one another, revise, run neutral tests, compete, and safely apply the winner. Each agent works in an isolated Git worktree, making every run reproducible and auditable. Open source, with 40 multi-agent LLM paper workflows included.
Claude Code, Codex, and other coding agents have different strengths. However, naive multi-agent orchestration such as simply launching several agents in parallel or allowing them to exchange messages does not define a reproducible development process. A real workflow must specify:
who implements;
who reviews whom;
when an agent must revise its work;
which candidates are independently tested;
how the winner is selected; and
when the selected change is applied to the original branch.
h5i-python is the Python SDK for the h5i orchestra engine. This SDK lets you define and execute multi-agent coding workflows across Claude Code, Codex, and other runtimes as ordinary Python programs.
Each agent works inside its own sandboxed Git worktree, so it cannot overwrite the original checkout or another agent's work. Agent turns produce Git-backed artifacts that can be reviewed, revised, neutrally verified, compared, selected, and applied as one auditable workflow.
Install
Quickstart
Create ensemble.py inside the Git repository the agents should modify. This workflow let Claude and Codex independently implement the same task, review and improve each other’s work, and then select the better result.
Run it as a normal Python program:
With the default launcher="resident", h5i automatically starts the agent sessions through tmux.
A dashboard view of each agent's live status and which worktree they're touching would be really helpful when you have several running at once. Right now it's hard to tell at a glance who is implementing what versus reviewing, especially when a disagreement turns into a few rounds of revisions.
Honestly this looks really cool, especially the part about agents reviewing each other in isolated worktrees. One thing I'd love to see is a built-in dashboard or CLI command that shows the current status of all running agents in real time, basically like a live view of which agent is doing what, what's queued, and which worktrees are active. Would make debugging multi-agent runs way less painful when something stalls.