Agent skill

Code Task Generator

by mikeyobrien in mikeyobrien/rho

Generate structured .code-task.md files from rough descriptions or PDD plans.

MITAuto-check passed

Install Code Task Generator

skills CLI
$ npx skills add mikeyobrien/rho --skill code-task-generator -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install mikeyobrien/rho code-task-generator --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/mikeyobrien/rho.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/code-task-generator .claude/skills/code-task-generator && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
code-task-generator
GitHub stars
372
Token cost
~3.3k tokens
SKILL.md length
1,193 words
Files
1
Skills in repo
35
Repo updated
First seen
Licence
MIT

At a glance

Generate structured .code-task.md files from rough descriptions or PDD plans.

  • Works in 6 steps: Detect Input Mode → Analyze Input → Structure Requirements → …
  • SKILL.md covers Overview, Parameters, Steps and Code Task Format Specification, plus 2 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Code Task Generator is an agent skill from mikeyobrien/rho. Generate structured .code-task.md files from rough descriptions or PDD plans.

Its SKILL.md is about 3.3k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

The repository describes itself as: An AI agent that stays running, remembers across sessions, and checks in on its own. macOS, Linux, Android. Built on Pi. The licence is MIT.

Example prompts

  • “/code-task-generator”

Workflow steps

6 steps, taken from the step headings in SKILL.md.

  1. Detect Input Mode
  2. Analyze Input
  3. Structure Requirements
  4. Plan Tasks
  5. Generate Tasks
  6. Report Results

What it can do on your machine

Read from SKILL.md and the folder at commit 073a3ee. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are markdown).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Code Task Generator loads about 3.3k tokens when it runs. Until then it costs about 24 tokens; SKILL.md has 1,193 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~24
When it runs · the whole SKILL.md, loaded when a task matches
~3.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from mikeyobrien/rho at commit 073a3ee, republished under its MIT licence (© mikeyobrien). 1,193 words, ~3,343 tokens.

Download SKILL.mdSave it as .claude/skills/code-task-generator/SKILL.md (or your agent's skills folder).
name
code-task-generator
description
Generate structured .code-task.md files from rough descriptions or PDD plans.
kind
sop

Code Task Generator

Overview

This sop generates structured code task files from rough descriptions, ideas, or PDD implementation plans. It automatically detects the input type and creates properly formatted code task files following Amazon's code task format specification. For PDD plans, it processes implementation steps one at a time to allow for learning and adaptation between steps.

Parameters

  • input (required): Task description, file path, or PDD plan path. Can be a simple sentence, paragraph, detailed explanation, or path to a PDD implementation plan.
  • step_number (optional): For PDD plans only - specific step to process. If not provided, automatically determines the next uncompleted step from the checklist.
  • output_dir (optional, default: ".agents/tasks/{project_name}"): Directory where the code task file will be created
  • project_name (optional): Project name for organizing tasks. If processing a PDD plan, will be inferred from the plan path. Otherwise, will be generated from the description with a YYYY-MM-DD date prefix.

Constraints for parameter acquisition:

  • You MUST ask for all required parameters upfront in a single prompt rather than one at a time
  • You MUST support multiple input methods for input including:
    • Direct text input
    • File path containing the description or PDD plan
    • Directory path (will look for plan.md within it)
    • URL to internal documentation
  • You MUST confirm successful acquisition of all parameters before proceeding

Steps

1. Detect Input Mode

Automatically determine whether input is a description or PDD plan.

Constraints:

  • You MUST check if input is a file path that exists
  • If file exists, You MUST read it and check for PDD plan structure (checklist, numbered steps)
  • If file contains PDD checklist format, You MUST set mode to "pdd"
  • If input is text or file without PDD structure, You MUST set mode to "description"
  • You MUST inform user which mode was detected
  • You MUST validate that PDD plans follow expected format with numbered steps
2. Analyze Input

Parse and understand the input content based on detected mode.

Constraints:

  • For PDD mode: You MUST parse implementation plan and extract steps/checklist status
  • For PDD mode: You MUST determine target step based on step_number parameter or first uncompleted step
  • For description mode: You MUST identify the core functionality being requested
  • You MUST extract any technical requirements, constraints, or preferences mentioned
  • You MUST determine the appropriate complexity level (Low/Medium/High)
  • You MUST identify the likely technology stack or domain area
3. Structure Requirements

Organize requirements and determine task breakdown based on mode.

Constraints:

  • For PDD mode: You MUST extract target step's title, description, demo requirements, and constraints
  • For PDD mode: You MUST preserve integration notes with previous steps
  • For PDD mode: You MUST identify which specific research documents (if any) are directly relevant to each task being created
  • For description mode: You MUST identify specific functional requirements from the description
  • You MUST infer reasonable technical constraints and dependencies
  • You MUST create measurable acceptance criteria using Given-When-Then format
  • You MUST prepare task breakdown plan for approval
4. Plan Tasks

Present task breakdown for user approval before generation.

Constraints:

  • You MUST analyze content to identify logical sub-tasks for implementation
  • You MUST present concise one-line summary for each planned code task
  • You MUST show proposed task sequence and dependencies
  • You MUST ask user to approve the plan before proceeding
  • You MUST allow user to request modifications to the task breakdown
  • You MUST NOT proceed to generate actual code task files until user explicitly approves
5. Generate Tasks

Create appropriate file structure based on mode and approved plan.

Constraints:

  • For PDD mode: You MUST create a folder named step{NN} where NN is zero-padded (e.g., step01, step02, step10)
  • For PDD mode: You MUST create multiple code task files within the step folder, named sequentially: task-01-{title}.code-task.md, task-02-{title}.code-task.md, etc.
  • For PDD mode: You MUST break down the step into logical implementation phases focusing on functional components, NOT separate testing tasks
  • For PDD mode: You MUST include "Reference Documentation" section with path to design/detailed-design.md as required reading
  • For PDD mode: You MUST include specific research documents in "Additional References" only if they are directly relevant to the task (e.g., specific technology research for that component)
  • For PDD mode: You MUST add a note instructing agents to read the detailed design before implementation
  • For description mode: You MUST create single task or multiple tasks as planned
  • You MUST generate task names using kebab-case format
  • You MUST create files with .code-task.md extension
  • You MUST follow the exact format specified in the Code Task Format section below
  • You MUST include comprehensive acceptance criteria that cover the main functionality
  • You MUST include unit test requirements as part of the acceptance criteria for each implementation task
  • You MUST NOT create separate tasks for "add unit tests" or "write tests" because testing should be integrated into each functional implementation task
  • You MUST provide realistic complexity assessment and required skills
  • You MUST save files to the specified output directory
Show full SKILL.md (395 more words)Show less
6. Report Results

Inform user about generated tasks and next steps.

Constraints:

  • You MUST list all generated code task files with their paths
  • For PDD mode: You MUST provide the step demo requirements for context
  • For description mode: You MUST provide a brief summary of what was created
  • You MUST suggest running code-assist on each task in appropriate sequence
  • For PDD mode: You MUST NOT create any additional log files or summary documents
  • For description mode: You MUST offer to create additional related tasks if the scope seems large

Code Task Format Specification

Each code task file MUST follow this exact structure:

markdown
# Task: [Task Name]

## Description
[A clear description of what needs to be implemented and why]

## Background
[Relevant context and background information needed to understand the task]

## Reference Documentation
**Required:**
- Design: [path to detailed design document]

**Additional References (if relevant to this task):**
- [Specific research document or section]

**Note:** You MUST read the detailed design document before beginning implementation. Read additional references as needed for context.

## Technical Requirements
1. [First requirement]
2. [Second requirement]
3. [Third requirement]

## Dependencies
- [First dependency with details]
- [Second dependency with details]

## Implementation Approach
1. [First implementation step or approach]
2. [Second implementation step or approach]

## Acceptance Criteria

1. **[Criterion Name]**
   - Given [precondition]
   - When [action]
   - Then [expected result]

2. **[Another Criterion]**
   - Given [precondition]
   - When [action]
   - Then [expected result]

## Metadata
- **Complexity**: [Low/Medium/High]
- **Labels**: [Comma-separated list of labels]
- **Required Skills**: [Skills needed for implementation]
Code Task Format Example
markdown
# Task: Create Email Validator Function

## Description
Create a function that validates email addresses and returns detailed error messages for invalid formats. This will be used across the application to ensure data quality and provide user-friendly feedback.

## Background
The application currently accepts any string as an email address, leading to data quality issues and failed communications. We need a robust validation function that can identify common email format errors and provide specific feedback to users.

## Reference Documentation
**Required:**
- Design: planning/design/detailed-design.md

**Additional References (if relevant to this task):**
- planning/research/validation-libraries.md (for email validation approach)

**Note:** You MUST read the detailed design document before beginning implementation. Read additional references as needed for context.

## Technical Requirements
1. Create a function that accepts an email string and returns validation results
2. Implement comprehensive email format validation using regex or email parsing library
3. Return detailed error messages for specific validation failures
4. Support common email formats including international domains
5. Include performance optimization for high-volume validation

## Dependencies
- Email validation library or regex patterns
- Error handling framework for structured error responses
- Unit testing framework for comprehensive test coverage

## Implementation Approach
1. Research and select appropriate email validation approach (regex vs library)
2. Implement core validation logic with specific error categorization
3. Add comprehensive error messaging for different failure types
4. Optimize for performance if needed for high-volume scenarios

## Acceptance Criteria

1. **Valid Email Acceptance**
   - Given a properly formatted email address
   - When the validation function is called
   - Then the function returns success with no errors

2. **Invalid Format Detection**
   - Given an email with invalid format (missing @, invalid characters, etc.)
   - When the validation function is called
   - Then the function returns failure with specific error message

3. **Detailed Error Messages**
   - Given various types of invalid emails
   - When validation fails
   - Then specific error messages are returned (e.g., "Missing @ symbol", "Invalid domain format")

4. **Performance Requirements**
   - Given 1000 email validations
   - When executed in sequence
   - Then all validations complete within 1 second

5. **Unit Test Coverage**
   - Given the email validator implementation
   - When running the test suite
   - Then all validation scenarios have corresponding unit tests with >90% coverage

## Metadata
- **Complexity**: Low
- **Labels**: Validation, Email, Data Quality, Utility Function
- **Required Skills**: Regular expressions, email standards, unit testing

Examples

Example Input (Description Mode)
input: "I need a function that validates email addresses and returns detailed error messages"
output_dir: ".agents/tasks/my-project"
Example Output (Description Mode)
Detected mode: description

Generated code task: .agents/tasks/my-project/email-validator.code-task.md

Created task for email validation functionality with comprehensive acceptance criteria and implementation guidance.

Next steps: Run code-assist on the generated task to implement the solution.
Example Input (PDD Mode)
input: ".agents/planning/my-project/implementation/plan.md"
Example Output (PDD Mode)
Detected mode: pdd

Generated code tasks for step 2: .agents/tasks/my-project/step02/

Created tasks:
- task-01-create-data-models.code-task.md
- task-02-implement-validation.code-task.md  
- task-03-add-serialization.code-task.md

Next steps: Run code-assist on each task in sequence

Step demo: Working data models with validation that can create, validate, and serialize/deserialize data objects

Troubleshooting

Vague Description (Description Mode)

If the task description is too vague or unclear:

  • You SHOULD ask clarifying questions about specific requirements
  • You SHOULD suggest common patterns or approaches for the domain
  • You SHOULD create a basic task and offer to refine it based on feedback
Complex Description (Description Mode)

If the description suggests a very large or complex task:

  • You SHOULD suggest breaking it into multiple smaller tasks
  • You SHOULD focus on the core functionality for the initial task
  • You SHOULD offer to create additional related tasks
Missing Technical Details (Description Mode)

If technical implementation details are unclear:

  • You SHOULD make reasonable assumptions based on common practices
  • You SHOULD include multiple implementation approaches in the task
  • You SHOULD note areas where the user should make technical decisions
Plan File Not Found (PDD Mode)

If the specified plan file doesn't exist:

  • You SHOULD check if the path is a directory and look for plan.md within it
  • You SHOULD suggest common locations where PDD plans might be stored
  • You SHOULD validate the file path format and suggest corrections
Invalid Plan Format (PDD Mode)

If the plan doesn't follow expected PDD format:

  • You SHOULD identify what sections are missing or malformed
  • You SHOULD suggest running the PDD script to generate a proper plan
  • You SHOULD attempt to extract what information is available
No Uncompleted Steps (PDD Mode)

If all steps in the checklist are marked complete:

  • You SHOULD inform the user that all steps appear to be complete
  • You SHOULD ask if they want to generate a task for a specific step anyway
  • You SHOULD suggest reviewing the implementation plan for potential new steps

© mikeyobrien, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/code-task-generator of mikeyobrien/rho.

Open the folder on GitHubat commit 073a3ee

Compare with similar skills

Code Task Generator next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Code Task Generator compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Code Task Generator this skillmikeyobrien/rho372—~3.3kAutomated safety check: PassMIT
Generatealirezarezvani/claude-skills28k1 repos~1.1kAutomated safety check: PassMIT
Fal Generatenexu-io/open-design100k—~306Automated safety check: PassApache-2.0
Video Generationbytedance/deer-flow83k4 repos~1.4kAutomated safety check: PassMIT
Image Generationonyx-dot-app/onyx32k1 repos~1.7kAutomated safety check: PassCustom licence
Structured Image Generationbytedance/deer-flow83k5 repos~2.9kAutomated safety check: PassMIT

Similar skills

  • Generate

    alirezarezvani/claude-skills

    Generate Playwright tests. An agent skill from alirezarezvani/claude-skills.

    28k GitHub starsUsed in 1 repo~1.1k tokens
    Testing & QAAuto-check passed
  • Fal Generate

    nexu-io/open-design

    Generate images and videos using fal.ai AI models. An agent skill from nexu-io/open-design.

    100k GitHub stars~306 tokensUpdated today
    Media & CreativeAuto-check passed
  • Video Generation

    bytedance/deer-flow

    Generates short videos from a structured JSON prompt, optionally guided by a reference image used as the first or last frame.

    83k GitHub starsUsed in 4 repos~1.4k tokens
    Media & CreativeAuto-check passed
  • Image Generation

    onyx-dot-app/onyx

    Generate or edit raster images (photos, illustrations, textures, sprites, mockups, logos, infographics) using the workspace's configured image-generation provider via onyx-cli image.

    32k GitHub starsUsed in 1 repo~1.7k tokens
    Media & CreativeAuto-check passed
  • Structured Image Generation

    bytedance/deer-flow

    Turns an image request into a structured JSON prompt and runs a bundled Python script to generate the picture, optionally guided by reference images.

    83k GitHub starsUsed in 5 repos~2.9k tokens
    Media & CreativeAuto-check passed
  • Generate Nanobanana

    sickn33/agentic-awesome-skills

    Generate and edit images/video with Google's Gemini media models (Nano Banana 2/Pro, Gemini Omni Flash), with cost-approval gates, reference-image support, and a prompt/output log per call.

    47k GitHub starsUsed in 1 repo~2.8k tokens
    Media & CreativeAuto-check: notes

More from mikeyobrien/rho

All 35 skills in this repo
  • Rho Cloud Onboard

    mikeyobrien/rho

    Register an agent email address on Rhobot Mail (name@rhobot.dev).

    372 GitHub stars~1.4k tokensUpdated 7 days ago
    Auto-check passed
  • Install Rho

    mikeyobrien/rho

    Install and configure Rho from scratch (Doom-style init.toml + sync).

    372 GitHub stars~1.6k tokensUpdated 7 days ago
    Auto-check: notes
  • Open URL

    mikeyobrien/rho

    Open URLs and launch apps on Android. An agent skill from mikeyobrien/rho.

    372 GitHub stars~480 tokensUpdated 7 days ago
    Auto-check passed
  • Release Changelog

    mikeyobrien/rho

    Keep CHANGELOG.md idiomatic (Keep a Changelog) and cut a tag-based GitHub release that triggers npm publish CI.

    372 GitHub stars~1.4k tokensUpdated 7 days ago
    Auto-check passed
  • Rho Cloud Email

    mikeyobrien/rho

    Manage agent email at name@rhobot.dev via the Rhobot Mail API.

    372 GitHub stars~2.6k tokensUpdated 7 days ago
    Auto-check passed
  • Tasker XML

    mikeyobrien/rho

    Create Tasker profiles and tasks via XML for Android automation.

    372 GitHub stars~3.1k tokensUpdated 7 days ago
    Auto-check passed

Questions about Code Task Generator

What does Code Task Generator do?

Generate structured .code-task.md files from rough descriptions or PDD plans. Code Task Generator is an agent skill from mikeyobrien/rho.md files from rough descriptions or PDD plans.

How do I install Code Task Generator in Claude Code?

Run `npx skills add mikeyobrien/rho --skill code-task-generator -a claude-code`. Or copy the skill folder (skills/code-task-generator in mikeyobrien/rho) into .claude/skills/code-task-generator in your project. Claude Code loads it when a task matches its description.

How do I install Code Task Generator in Codex?

Run `npx skills add mikeyobrien/rho --skill code-task-generator -a codex`. Or copy the skill folder (skills/code-task-generator in mikeyobrien/rho) into .agents/skills/code-task-generator in your project. Codex loads it when a task matches its description.

Can I use Code Task Generator in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add mikeyobrien/rho --skill code-task-generator -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/code-task-generator, .gemini/skills/code-task-generator, .github/skills/code-task-generator and .opencode/skills/code-task-generator in your project.

What does Code Task Generator need to run?

SKILL.md names no scripts, command-line tools or credentials: Code Task Generator is instructions for the agent only.

Does Code Task Generator access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Code Task Generator safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Code Task Generator use?

Code Task Generator is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Code Task Generator use?

About 3.3k tokens (SKILL.md is roughly 13k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Code Task Generator?

Skills that share tags, products or a category with Code Task Generator: Generate (alirezarezvani/claude-skills, 28k stars), Fal Generate (nexu-io/open-design, 100k stars), Video Generation (bytedance/deer-flow, 83k stars) and Image Generation (onyx-dot-app/onyx, 32k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Code Task Generator?

mikeyobrien (a GitHub user) maintains it in mikeyobrien/rho, which has 372 GitHub stars. The repository holds 35 skills in this directory. The repository was last updated on October 1, 2026.

Source: mikeyobrien/rho on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.