code-task-generator — Claude Code skill: tasks and setup
Turn a feature description or PDD plan into .code-task.md files with acceptance criteria. The skill shows its proposed task breakdown before writing anything.
npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a claude-codeWhat code-task-generator does
Give the skill a task description or a PDD plan path. It identifies the input mode, extracts requirements, and drafts Given-When-Then acceptance criteria that include unit testing. First, it presents the task sequence and dependencies. It writes files only after you approve the plan.
With a PDD plan, the skill works through one step at a time and makes the design document required reading. By default, task files go under .ralph/tasks/; PDD tasks are grouped in a step folder and named in sequence. Input can be direct text, a file or directory path, or a URL.
PDD tasks reference .ralph/specs/{task_name}/design.md as required documentation. Check that this path matches your project, since the generated task expects the design document to be there.
What it helps with
- split a feature description into implementation tasks
- turn a PDD step into an ordered set of task files
- write testable Given-When-Then acceptance criteria
- include unit test requirements in each task
Example prompts for the agent
- Create tasks for a feature that validates email addresses and returns detailed error messages.
- Read
.ralph/specs/data-pipeline/plan.mdand prepare tasks for its next incomplete step. - Draft a task breakdown for this description and wait for my approval before creating files.
How to install code-task-generator
1Skills CLI (Vercel)
npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a claude-code2Manually: put the files in place
~/.claude/skills/code-task-generator/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/code-task-generator/
1Skills CLI (Vercel)
npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a codex2Manually: put the files in place
.agents/skills/code-task-generator/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a cursor2Manually: put the files in place
.cursor/skills/code-task-generator/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install mikeyobrien/ralph-orchestrator code-task-generator2Manually: put the files in place
.github/skills/code-task-generator/SKILL.md1Skills CLI (Vercel)
npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a gemini-cli2Platform CLI
gemini skills install https://github.com/mikeyobrien/ralph-orchestrator --path .claude/skills/code-task-generatorManually: put the files in place
.devin/skills/code-task-generator/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about code-task-generator
How do I install code-task-generator in Claude Code?
Run npx skills add mikeyobrien/ralph-orchestrator --skill code-task-generator -a claude-code. The skill file is at ~/.claude/skills/code-task-generator/SKILL.md.
What does code-task-generator do?
It creates .code-task.md files from descriptions or PDD plans, with acceptance criteria and unit test requirements. It asks you to approve the breakdown before creating files.
How does the skill handle PDD plans?
It processes the selected step, or the first unfinished one if no step is specified. Each step is handled separately, and its task files reference the design document.
Source text
SKILL.md · MIT
---
name: code-task-generator
description: Generates structured .code-task.md files from descriptions or PDD implementation plans. Auto-detects input type, creates properly formatted tasks with Given-When-Then acceptance criteria.
type: anthropic-skill
version: "1.1"
metadata:
internal: true
---
# Code Task Generator
## Overview
Generate structured code task files from rough descriptions or PDD implementation plans. Auto-detects input type and creates properly formatted `.code-task.md` files. For PDD plans, processes one step at a time to allow learning between steps.
## Important Notes
These rules apply across ALL steps:
- **User approval required:** Present the task breakdown plan and get explicit approval before generating any files.
- **Tests are integrated:** Include unit test requirements in each task's acceptance criteria. Never create separate "add tests" tasks.
- **PDD mode references:** Always include the design document path as required reading. Only include research docs if directly relevant to the specific task.
## Parameters
- **input** (required): Task description, file path, or PDD plan path
- **step_number** (optional, PDD only): Specific step to process. Auto-determines next uncompleted step if omitted.
- **output_dir** (optional, default: `.ralph/tasks/`): Output directory for code task files
- **task_name** (optional, description mode only): Override auto-generated task name
**Constraints:**
- You MUST ask for all required parameters upfront in a single prompt
- You MUST support input as: direct text, file path, directory path (looks for plan.md), or URL
## Steps
### 1. Detect Input Mode
Check if input is a file with PDD plan structure (checklist + numbered steps). Set mode to "pdd" or "description" and inform the user.
### 2. Analyze Input
- **PDD mode:** Parse the plan, extract steps and checklist status, determine target step (from step_number or first uncompleted)
- **Description mode:** Identify core functionality, technical requirements, complexity level (Low/Medium/High), and technology domain
### 3. Structure Requirements
- **PDD mode:** Extract the target step's title, description, demo requirements, constraints, and integration notes with previous steps. Identify relevant research documents.
- **Description mode:** Identify functional requirements, infer technical constraints and dependencies.
For both modes: create measurable acceptance criteria in Given-When-Then format and prepare a task breakdown plan.
### 4. Plan Tasks
Present the proposed breakdown to the user:
- One-line summary per task
- Proposed sequence and dependencies
- You MUST NOT generate files until the user explicitly approves
### 5. Generate Tasks
Create files following the Code Task Format below.
**PDD mode specifics:**
- Create `.ralph/tasks/{task_name}/step{NN}/` folder by default (zero-padded: step01, step02, step10)
- Name files sequentially: `task-01-{title}.code-task.md`, `task-02-{title}.code-task.md`
- Break down by functional components, not testing phases
**All tasks:**
- You MUST use the exact Code Task Format structure below
- You MUST include YAML frontmatter with `status: pending`, `created: YYYY-MM-DD`, `started: null`, `completed: null`
- You MUST use kebab-case names with `.code-task.md` extension
- You MUST include acceptance criteria covering main functionality and unit tests
### 6. Report Results
List generated files with paths. For PDD mode, include the step's demo requirements. Suggest running code-assist on tasks in sequence, or using Ralph for autonomous implementation.
### 7. Offer Ralph Integration
Ask: "Would you like me to set up Ralph to implement these tasks autonomously?"
If yes, create a concise PROMPT.md with objective, spec directory reference, execution order, and acceptance criteria. Suggest the appropriate command:
- Full pipeline: `ralph run --config presets/pdd-to-code-assist.yml`
- Simpler flow: `ralph run -c ralph.yml -H builtin:code-assist`
## Code Task Format Specification
Each code task file MUST follow this structure:
```markdown
---
status: pending
created: YYYY-MM-DD
started: null
completed: null
---
# Task: [Task Name]
## Description
[What needs to be implemented and why]
## Background
[Context needed to understand the task]
## Reference Documentation
**Required:**
- Design: .ralph/specs/{task_name}/design.md
**Additional References (if relevant to this task):**
- [Specific research document or section]
**Note:** Read the design document before beginning implementation.
## Technical Requirements
1. [First requirement]
2. [Second requirement]
## Dependencies
- [Dependency with details]
## Implementation Approach
1. [Implementation step or approach]
## Acceptance Criteria
1. **[Criterion Name]**
- Given [precondition]
- When [action]
- Then [expected result]
## Metadata
- **Complexity**: [Low/Medium/High]
- **Labels**: [Comma-separated labels]
- **Required Skills**: [Skills needed]
```
## Examples
**Description mode input:** `"I need a function that validates email addresses and returns detailed error messages"`
**Description mode output:** `.ralph/tasks/email-validator.code-task.md` — task with acceptance criteria for valid/invalid email handling, error messages, and unit tests.
**PDD mode input:** `".ralph/specs/data-pipeline/plan.md"`
**PDD mode output:** `.ralph/tasks/data-pipeline/step02/` containing `task-01-create-data-models.code-task.md`, `task-02-implement-validation.code-task.md`, `task-03-add-serialization.code-task.md` — each with design.md reference, acceptance criteria, and demo requirements.
## Troubleshooting
**Vague description:** Ask clarifying questions, suggest common patterns, create a basic task and offer to refine.
**Complex description:** Suggest breaking into smaller tasks, focus on core functionality first, offer to create related tasks.
**Missing technical details:** Make reasonable assumptions, include multiple approaches, note areas needing user decisions.
**Plan file not found:** Check if path is a directory (look for plan.md within), suggest common PDD plan locations.
**Invalid plan format:** Identify missing sections, suggest running PDD to generate a proper plan, extract what's available.
**All steps complete:** Inform user, ask if they want a specific step anyway, suggest reviewing for new steps.
© mikeyobrien · MIT · mikeyobrien/ralph-orchestrator