Back to Ecosystem Pulse
OperationsExperimental
agent-manager-skill
by fractalmind-ai
Serverless tmux-based agent lifecycle manager with cron scheduling
Python
Updated Aug 19, 2026
Share:
Overview
Manages agent lifecycles locally using tmux and a Python CLI to start, stop, monitor, and assign agents without a server. Uses cron-friendly scheduling and lightweight process supervision so you can run many agents on a single host with simple scripts. Distinctive for its minimal, no-server approach and tmux-based session tracking for easy debugging and session recovery. tmux-based session tracking
Key Benefits
As multi-agent deployments grow, simple process-level tooling is often missing from evaluation and trust workflows — you need reliable start/stop and consistent logs to build agent track records. This repo fills that gap by providing repeatable lifecycle controls and scheduling that make it easier to capture agent interaction logs and failure modes. That basic operational hygiene is a prerequisite for meaningful agent-to-agent evaluation and reputation signals. Evaluation-Driven Development (EDDOps)
Ideal For
Developers running local or single-host multi-agent experiments who want lightweight lifecycle control and reproducible scheduling without adding orchestration infrastructure. orchestration Tool Use Pattern
Real-World Examples
- Run and schedule multiple agents on a single host with cron-friendly jobs
- Capture consistent session logs and agent failure modes for evaluation
- Quickly start/stop and reassign agents during iterative development and debugging
Works With
tmuxcronpythoncli
Topics
agent-managementagent-orchestrationai-agentai-agentsclicronfractalmulti-agentopenskillsorchestration+2 more
Keywords
multi-agent orchestrationproduction agent monitoringagent lifecycletmux