skills.rest
silcam
silcamOFFICIAL
Explore all skills by this author →
GitHub

ralph

Drain ready task queues and manage task bookkeeping for Claude Code sessions.

2|Updated Mar 27, 2019
Search Tags:#git workflow#claude code#beads#subagent#task orchestration#queue draining
One-click install
npx skills add https://github.com/silcam/lessons-from-luke --skill ralph-silcam
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/silcam/lessons-from-luke/tree/main/.claude/skills/ralph
Command: npx skills add https://github.com/silcam/lessons-from-luke --skill ralph-silcam

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralph removes the manual overhead of draining ready task queues, claiming the right leaf task, and confirming completion without relying on a separate daemon or ad hoc orchestration.

Core Features & Use Cases

  • Runs prep and verify bookkeeping inline in the main session so state changes persist reliably.
  • Dispatches one short-lived worker per leaf task to keep implementation work isolated.
  • Supports epic-scoped draining from the current branch or repository-wide draining when no epic matches.
  • Handles failure recovery, queue-empty detection, and hot-loop protection.
  • Use case: a developer on a feature branch can let the skill process every ready leaf under the matching epic while keeping the session deterministic.

Quick Start

Tell the assistant to drain the current ready queue for this repository and report each task as it is claimed, worked, and verified.

Frequently Asked Questions about ralph

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

FAQPage Schema
How do I automate draining ready task queues for Claude Code?▼

You can automate draining ready task queues by running inline prep and verify bookkeeping, dispatching single-worker subagents per leaf task, and repeating claim-implement-verify cycles until the queue is empty.

What is task bookkeeping in autonomous Claude Code session orchestration?▼

Task bookkeeping tracks state changes inline during the main session, ensuring that prep and verify steps persist reliably while short-lived workers execute isolated leaf-level implementation work.

How do I handle failed or stalled tasks during queue draining?▼

Failed or stalled tasks are handled through built-in failure recovery and hot-loop protection, which safely detects stalled work and advances HEAD-advance checks to maintain deterministic session state.

Can I drain task queues repository-wide instead of just for a single epic?▼

Yes, you can drain queues repository-wide when no matching epic is found, or restrict draining to epic-scoped ready queues based on the current feature branch.

Do I need a separate daemon to orchestrate Claude Code task queues?▼

No, you do not need a separate daemon or ad hoc orchestration; the skill autonomously orchestrates the queue draining process entirely within the main Claude Code session.

What are the limitations of using inline subagent dispatch for task orchestration?▼

The skill dispatches only one short-lived worker per leaf task to keep work isolated, meaning concurrent execution is limited and queue draining is strictly sequential rather than parallel.

Related Skills

Explore More Software Engineering Skills→
usebonderyusebondery
15

pm-feature-refiner

Decompose Plane features into implementation-ready subtasks with acceptance criteria.

Official
Advanced
apokamoapokamo
12

issue-verify-code

Verify code fixes against prior feedback and git diff HEAD~1.

Community
Intermediate
xxihxxih
1

subagent-driven-development

Dispatch fresh subagents for isolated task execution with two-stage reviews.

Community
Advanced
GregoryHoGregoryHo
6

arc-dispatching-parallel

Dispatch independent tasks in parallel using DAG dependency structures.

Community
Advanced