long-horizon

Name: long-horizon
Availability: InStock
Author: Ryan

Orchestrates long multi-step tasks across sessions using verified executor and auditor rounds.

Updated Aug 30, 2026

Search Tags:#context management #audit #workflow orchestration #task automation #multi-agent #long-running tasks #state persistence

Ready to run

npx skills add https://github.com/ryanportfolio/threejs-interview-test --skill long-horizon-ryanportfolio

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: long-horizon
Source: https://github.com/ryanportfolio/threejs-interview-test/tree/main/.agents/skills/long-horizon
Command: npx skills add https://github.com/ryanportfolio/threejs-interview-test --skill long-horizon-ryanportfolio

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Long tasks that exceed a single context window lose progress to compaction, failed retries, or session boundaries, forcing work to restart from scratch. ## Core Features & Use Cases - Verified Rounds: Runs work in executor and auditor rounds so each step is independently checked before continuing. - State Persistence: Tracks progress through state and workspace files so work survives compaction and multi-hour or multi-session execution. - Use Case: When asked to complete a large refactoring or multi-phase build, invoke the skill to break it into verified rounds, spawn fresh executor agents with standalone briefs, and audit each round before proceeding. ## Quick Start Ask the agent to run this large task in verified rounds using the long-horizon workflow.

Frequently Asked Questions about long-horizon

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

FAQPage Schema

How do I run a long task that exceeds one context window?▼

Break the work into verified rounds where an executor agent performs each step and an independent auditor checks the result. Persist progress in state and workspace files so later rounds resume without relying on conversation history.

How do I recover progress lost to context compaction?▼

Store task state in files rather than chat history, then spawn fresh agents with standalone briefs pointing to those state and workspace paths. This lets work continue after compaction or failed retries without re-explaining the task.

Can executor agents see previous conversation turns?▼

By default executor and auditor rounds spawn agents with fork_turns set to none, receiving only a standalone brief plus state paths. A positive turn count is an exception only when required data cannot be recovered from files, and must be recorded in the audit log.

Does invoking the skill authorize commits or deployments?▼

No. Skill invocation alone does not authorize commits, pushes, pull requests, merges, deploys, migrations, or dependency installation. Those outward actions require an explicit user request.

When should I not use a long-horizon workflow?▼

Avoid it for short single-step tasks that fit comfortably in one context window, since round-based execution and auditing add overhead. It is designed for work spanning hours, sessions, or many dependent steps.