autonomous-skill

Automate multi-session long-running tasks with a dual-agent pattern.

1|Updated Mar 8, 2026

Search Tags:#automation #multi-session #claude-code #task-management #autonomous #long-running-tasks

Ready to run

npx skills add https://github.com/tense-i/stock-market-simulator --skill autonomous-skill-tense-i

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: autonomous-skill
Source: https://github.com/tense-i/stock-market-simulator/tree/main/skills/autonomous-skill
Command: npx skills add https://github.com/tense-i/stock-market-simulator --skill autonomous-skill-tense-i

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires claude, and includes scripts (resource) components.

What problem does it solve?

Long-running tasks across multiple sessions are hard to coordinate, track, and resume. This skill introduces a dual-agent pattern (Initializer + Executor) to decompose work, maintain progress, and autonomously continue sessions using Claude Code in headless mode.

Core Features & Use Cases

Task decomposition and session orchestration: Breaks complex requests into manageable sub-tasks and runs repeated Claude Code sessions until completion.
Progress tracking and isolation: Creates per-task directories with task_list.md and progress.md to ensure clean, isolated execution state.
Automatic continuation: Automatically resumes subsequent sessions with controlled delays until all tasks are complete, enabling long-running automation in a robust way.

Quick Start

Tell the system to start a long-running automation and let it manage sessions to completion.

Frequently Asked Questions about autonomous-skill

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

FAQPage Schema

How do I automate long-running tasks across multiple Claude Code sessions?▼

Multi-session task automation uses a dual-agent pattern to decompose work and automatically resume Claude Code headless sessions until completion. It tracks state via task_list.md and progress.md in isolated task directories.

What is the dual-agent pattern for autonomous task execution?▼

The dual-agent pattern uses an Initializer to decompose complex requests into sub-tasks and an Executor to run them. It maintains cross-session continuity by updating progress.md and automatically triggering subsequent sessions.

How do I track progress and maintain state for cross-session automation?▼

Cross-session progress tracking is handled by creating per-task directories containing task_list.md and progress.md. These files are automatically updated as work progresses to ensure clean, isolated execution state.

Do I need Claude Code headless mode to run autonomous multi-session tasks?▼

Yes, Claude Code headless mode is required for automatic session continuation. It enables the system to autonomously resume subsequent sessions with controlled delays until all decomposed tasks are complete.

What's the best way to resume interrupted long-running automation tasks?▼

Automatic continuation resumes interrupted tasks by reading the saved state from progress.md and task_list.md. It triggers repeated headless sessions with controlled delays to robustly complete remaining sub-tasks without manual intervention.

Can I include custom scripts and references when orchestrating multi-session tasks?▼

Yes, multi-session task orchestration supports optional scripts, references, and assets as needed. These are managed within the isolated task directories alongside the standard progress tracking files.