auto

Orchestrate multi-step agentic workflows with a persistent state machine.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/shawnroos/auto --skill auto-shawnroos
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: auto
Source: https://github.com/shawnroos/auto/tree/main/skills/auto
Command: npx skills add https://github.com/shawnroos/auto --skill auto-shawnroos

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill solves the problem of fragile, manual-heavy agent workflows by providing a persistent, observable state machine that manages complex plan-build-review-fix loops.

Core Features & Use Cases

  • Durable Execution: Maintains a disk-persisted run-record that survives session exits and rate limits, allowing for seamless resumption.
  • Workflow Orchestration: Drives any toolchain through a pulse-based loop, managing fan-out, verification, and stalled-node recovery automatically.
  • Use Case: Use this to manage a multi-step software engineering task where you need to plan, implement, and review code changes across multiple files without manually babysitting the agent through every iteration.

Quick Start

Start an auto run in the current repository to drive the default plan-build-review loop until the exit predicate is met.

Frequently Asked Questions about auto

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate complex multi-step agent workflows that survive rate limits and session disconnects?

To automate complex agent workflows durably, you need a state-machine driven loop engine that maintains a disk-persisted run-record. This allows your orchestration process to survive session exits and resume seamlessly without losing execution state.

What is the best way to orchestrate plan, build, and review loops across a software engineering toolchain?

Orchestrating plan-build-review loops requires a pulse-based state machine that manages fan-out and verification automatically. This approach drives your toolchain through iterative code changes without requiring manual intervention during each execution cycle.

How does watchdog-based stall detection work in agentic workflow orchestration?

Watchdog-based stall detection monitors the persistent state machine for stalled nodes during execution. When an active workflow node stops responding, the watchdog automatically triggers recovery mechanisms to resume the orchestration loop without manual babysitting.

Can I implement automated human-in-the-loop handoffs in a long-running automation workflow?

Yes, automated human-in-the-loop handoffs are supported by the durable state machine during long-running automation workflows. The orchestration engine pauses execution and routes the persistent run-record to a human for review before resuming the loop.

Do I need external dependencies to manage atomic run-record updates for fault-tolerant automation?

No external dependencies are required to manage atomic run-record updates for fault-tolerant automation. The state-machine loop engine handles disk persistence and atomic updates internally, ensuring consistent workflow state across toolchains.

When should I avoid using a state-machine loop engine for workflow automation?

You should avoid using a state-machine loop engine if your workflow automation requires only simple, single-step execution without fault tolerance. The overhead of maintaining a persistent run-record and watchdog monitoring is unnecessary for non-iterative, short-lived tasks.