worker-agents

Spawn isolated Claude Code sessions for parallel long-running tasks.

88|108|Updated Apr 7, 2026
One-click install
npx skills add https://github.com/grandamenium/cortextos --skill worker-agents-grandamenium
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: worker-agents
Source: https://github.com/grandamenium/cortextos/tree/main/templates/analyst/.claude/skills/worker-agents
Command: npx skills add https://github.com/grandamenium/cortextos --skill worker-agents-grandamenium

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It prevents slow, long-running work from blocking your main agent context by running independent tasks in separate ephemeral Claude Code sessions.

Core Features & Use Cases

  • Spawn isolated worker sessions for parallel work that needs a full Claude Code environment (tools, memory, context) rather than a lightweight subagent call.
  • Bus-based coordination to monitor progress and collect deliverables via the shared file bus between parent and workers.
  • Safety guardrails by scoping tasks and enforcing “no overlapping file sets” to avoid conflicts with other active agents.
  • Use Case: Run overnight research/design or multi-part drafting (e.g., separate experiments, doc drafts, or scaffolding) while your persistent Orchestrator continues handling approvals and other work.

Quick Start

Use worker-agents to spawn an isolated worker that creates the required deliverables in a dedicated directory and then reports completion back to your parent via the bus.

Frequently Asked Questions about worker-agents

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

FAQPage Schema
How do I prevent long-running Claude Code tasks from blocking my main context?

Parallelize tasks in isolated worker sessions to prevent context blocking. By spawning ephemeral sessions with scoped working directories, your parent agent can continue handling approvals while independent work runs separately.

Can I run parallel Claude Code sessions without file editing conflicts?

Yes, you can run parallel sessions safely by enforcing strict non-overlapping file guardrails. Scoping tasks to dedicated directories ensures separate active agents do not edit the same files concurrently during execution.

What is the best way to monitor progress of background research tasks spawned in parallel?

Monitor progress through a shared file bus communication mechanism. The parent session uses bus identity communication to track spawned workers, collect deliverables, and log spawn and completion events.

How do I orchestrate multi-part drafting across separate ephemeral worker sessions?

Orchestrate multi-part drafting by spawning isolated workers for each deliverable in a dedicated directory. Workers complete their drafting tasks independently and report completion back to the parent via the file bus.

When should I use isolated worker sessions instead of lightweight subagent calls?

Use isolated worker sessions for long-running parallel work requiring a full Claude Code environment with tools, memory, and context. Lightweight subagent calls are better suited for smaller, synchronous operations.

Does spawning parallel worker sessions require overlapping file guardrails for safe execution?

Spawning parallel workers requires strict non-overlapping file guardrails to ensure safe execution. This safety measure prevents conflicts and data corruption when multiple active agents operate concurrently.