Task Orchestrator

This is serious infrastructure for coordinating multiple Codex agents on large codebases. It spins up isolated git worktrees and tmux sessions per task, analyzes file dependencies to determine what can run in parallel versus what needs to serialize, and includes a heartbeat monitor that auto-recovers stuck sessions by sending input or restarting with error context. You'd reach for this when facing a backlog of 10+ GitHub issues that touch different parts of the system and you want autonomous parallel execution instead of babysitting each one. The self-healing is clever but also means you need to trust it with your repo. Works best when paired with the senior-engineering skill for better judgment calls.

GitHub stars: 235
last updated: Jun 10, 2026

View on GitHub

Install

$ npx -y skills add jdrhyne/agent-skills --skill task-orchestrator --agent claude-code

Installs into .claude/skills of the current project.

Files

Task Orchestrator

Autonomous orchestration of multi-agent builds using tmux + Codex with self-healing monitoring.

Load the senior-engineering skill alongside this one for engineering principles.

Safety Boundaries

Do not launch parallel workers for tasks with overlapping write scope until the dependency is resolved.
Do not push branches, merge work, or self-heal by guessing when human review is required.
Do not store secrets in manifests, logs, prompts, or tmux pane captures.
Do not continue retrying a failing task indefinitely; stop and surface the blocker after bounded retries.