Agent Skills: Generate Task (Worker Task Prompt)

<task_description>$ARGUMENTS</task_description>

Generate Task (Worker Task Prompt)

You generate ONE worker task prompt that will be executed by a worker agent.

You MUST follow the existing canonical task writing standard and structure:

• CLEAR ordering (Context, Objective, Inputs, Requirements, Constraints, Expected Outputs, Acceptance Criteria, Verification Steps, CoVe Checks only if needed, Handoff).
• Task structure requirements and fields (task, title, status, agent, dependencies, priority, complexity, accuracy-risk, parallelize-with, reason, handoff, Required Inputs).

Do NOT add any new fields, sections, agents, or mechanisms beyond what is already defined in the referenced task standards.

Inputs

When invoked, you will be given some combination of:

• a task title and brief description (in <task_description/>)
• optionally: dependencies, repo/file references, constraints, and verification expectations

If critical information is missing, you MUST keep the task executable by:

• stating assumptions under Required Inputs (and how to confirm them), and
• ensuring Verification Steps can confirm correctness, or explicitly indicate what blocks verification.

Output

Output exactly ONE task prompt in this format:

```yaml
---
task: [Task ID]
title: [Descriptive Name]
status: not-started
agent: [agent-name or "unassigned"]
dependencies: []
priority: [1-5 based on dependency depth]
complexity: [low/medium/high based on scope, not time]
accuracy-risk: [low/medium/high]
parallelize-with: []
reason: [Why parallelization is safe; avoid file conflicts]
handoff: [What the worker must report back: summary, evidence, blockers]
---
```

## Context
[Only what the worker needs; reference specific files/sections]

## Objective
[One sentence definition of success]

## Required Inputs
- [Files/links/artifacts the worker must read]
- [Assumptions and how to confirm them]

## Requirements
1. [Must do]
2. [Must do]

## Constraints
- [Must not do]
- [Guardrails, scope boundaries]

## Expected Outputs
- [Files created/modified with paths]
- [Artifacts produced]

## Acceptance Criteria
1. [Specific, measurable criterion]
2. [Another verifiable requirement]

## Verification Steps
1. [How to verify criterion 1]
2. [How to verify criterion 2]
N. (When Expected Outputs lists file paths) Run: `git add <file1> [file2 ...]` then
   `git commit -m "<type>(<scope>): <task title>"` — scope is the primary affected module or
   directory (required by repo commit-msg hook); use files from Expected Outputs only, no
   `git add .` or `git add -A`, no `Fixes #N` / `Closes #N` / `Resolves #N` trailer.

## CoVe Checks (ONLY if accuracy-risk is medium or high)
- Key claims to verify:
  - [Claim 1]
- Verification questions:
  1. [Question 1]
- Evidence to collect:
  - [Commands, docs, code pointers]
- Revision rule:
  - If any check fails, revise and state what changed.

Lint Before Final Output

Before returning the task prompt, you MUST lint it using the existing rules:

• Concise: no filler, no duplicated requirements
• Logical: sections in canonical order
• Explicit: objective, outputs, acceptance criteria, verification are concrete
• CoVe: included only when Accuracy Risk is Medium/High, and questions are falsifiable
• Commit step: when ## Expected Outputs lists one or more repo-relative file paths, ## Verification Steps MUST end with a git add <files> && git commit -m "<type>(<scope>): <title>" step scoped to those exact files — scope is required (repo enforces --force-scope), never git add . or git add -A, never a Fixes #N trailer

If any lint check fails, revise the task prompt and re-lint.