How does Brewcode handle tasks that exceed the context window?

It uses a sophisticated execution loop with automatic handoffs and a task registry to maintain state across multiple sessions and sub-agent invocations, effectively providing 'infinite' focus.

Does this skill work with any specific file structure?

It relies on a standardized directory structure including a PLAN.md, a phase registry, and an artifacts directory to track progress and manage sub-agent outputs.

Can I resume a task after it has been interrupted or compacted?

Yes, the skill is designed to resume seamlessly. It reads the current task state from the TaskList and PLAN.md to continue from the last pending or in-progress step.

What happens if a sub-agent fails during a task phase?

The orchestrator includes a built-in retry mechanism. If failures persist, it follows a defined escalation table that can include root-cause analysis, phase splitting, or model upgrades.

Brewcode Task Orchestrator

Name: Brewcode Task Orchestrator
Author: kochetkov-ma

bykochetkov-ma

•

Productivity & Workflow

Orchestrates complex development tasks with infinite context management and automated sub-agent handoff.

Brewcode Task Orchestrator is the core execution engine of the Brewcode platform, designed to manage long-running, multi-phase development projects within Claude Code. It automates the entire lifecycle of a task by parsing project plans, managing task dependencies, spawning specialized sub-agents for specific phases, and handling automatic context handoffs to overcome token limits. With built-in failure recovery, escalation protocols, and knowledge persistence, it ensures that complex software engineering tasks are completed systematically and reliably without losing track of project goals.

Key Features

01Dynamic task dependency mapping based on phase registries in PLAN.md

02Infinite context management via automated task handoff and state persistence

03Automated sub-agent spawning and coordination for specialized development roles

0416 GitHub stars

05Robust failure handling with multi-step retry logic and escalation paths

06Integrated knowledge management and automated project rule extraction

Use Cases

01Executing multi-step architectural migrations that exceed standard AI context windows

02Managing long-running refactoring projects with automated verification and fix loops

03Automating complex feature implementations involving multiple specialized sub-tasks and agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kochetkov-ma/claude-brewcode start

For use in Claude.ai and ChatGPT

Download Skill