new-task

SkillProductivity

Create an OpenMates Task for a concrete engineering outcome.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the new-task skill

What this skill tells your AI

The instructions your AI receives, as published by glowingkitty/openmates in .agents/skills/new-task/SKILL.md and read by ahel’s review.

Instructions

You are creating a new task for the OpenMates project. Your job is to take the user's raw input (which may be a brain dump, a bug report, a feature idea, or a vague note) and turn it into a well-structured task with all fields properly filled.

Use GitHub Issues by default. Use Linear only for programmatically stored/recorded product issues, marketing work, sensitive/private work, or when the user explicitly asks for Linear.

Step 1: Parse the Input

Analyze the user's input and determine:

  1. Type: Bug, Feature, Improvement, Research, Idea, DevOps, or Testing
  2. Title: Clean, concise title (max 80 chars). Rules:
    • Remove prefix stacking ("Fix: Fix:" → "Fix:")
    • Start with a verb for actionable tasks ("Fix", "Add", "Implement", "Research", "Investigate")
    • For ideas/questions, use plain language without prefixes
  3. Description: Expand the input into a structured description
  4. Priority: Suggest based on type and keywords:
    • Bug with "broken", "crash", "fails" → High (2)
    • Security issue → Urgent (1)
    • Feature request → Medium (3)
    • Nice-to-have, idea → Low (4) or None (0)
    • Research → Medium (3)
  5. Labels: Suggest from existing labels:
    • Bug, Feature, Improvement, Idea, Security, Performance, Encryption
    • UI, Backend, DevOps, testing, docs, marketing, payment
    • cronjob, user-feedback, SEO
  6. Project: Suggest Web App or CLI based on content
  7. Milestone: Suggest the current product line for ongoing work (for example v0.17) or the next minor product line for larger features (for example v0.17). Alpha artifact numbers such as 0.17.0-alpha.1 are package revisions, not user-facing milestones.
  8. Spec size: Decide none, inline, or full using docs/contributing/guides/spec-driven-development.md

Step 2: Present the Suggested Task

Show the user what will be created:

## New Task Preview

**Title:** Fix broken reminder creation UI
**Type:** Bug Fix
**Priority:** High (2)
**Labels:** Bug, UI
**Project:** Web App
**Milestone:** v0.17

### Description
The create reminder UI is broken — [expanded description based on user input].

### Acceptance Criteria
- [ ] Reminder creation works reliably
- [ ] UI matches Figma design

### Example Scenarios
- S-1: Given ..., when ..., then ...

### Spec Readiness
- Spec size: inline | full | none
- Full Plan path: docs/plans/<slug>/plan.yml (if needed)

Step 3: Ask for Confirmation

Use AskUserQuestion:

  • Create — create the task as shown
  • Edit — let the user adjust fields before creating

Step 4: Create the Task

For GitHub-default tasks, call github_issue_write with a concise title, structured body, labels, and assignees only when the user requested them.

If the task needs a full Plan, include a Plan Readiness section in the issue body and tell the user to run create-plan before implementation. Do not implement full-Plan work directly from a bare issue.

For Linear-only tasks, use python3 scripts/linear.py create --team OPE --title "..." --description "..." and add state/priority/labels when needed. Do not use Linear MCP tools.

If the task type is "Idea", automatically add the Idea label and set priority to None (0).

Step 5: Report

Show the created task with its identifier and URL:

Created: #123 — "Title"
URL: https://github.com/glowingkitty/OpenMates/issues/123

For Linear-only tasks, show the OPE-XXX identifier returned by scripts/linear.py.


Smart Suggestions

Title Cleanup Rules

  • "Fix: Fix Show More chats: pagination broken""Fix: Show More chats pagination broken"
  • "Feat: Feat: test auto-create""Feat: test auto-create linear issue"
  • "fix it" → Ask user to clarify. Every clarifying question must include Recommendation: with the evidence-based preferred answer and rationale plus Examples: with task-specific options or outcomes; if uncertain, recommend the safest reversible default and state the uncertainty.
  • Questions → Keep as-is but add Idea label if no clear action

Description Templates

Bug:

## Problem
[What's broken]

## Expected Behavior
[What should happen]

## Steps to Reproduce
1. ...

## Related
- Files: ...
- Commits: ...

## Spec Readiness
- Spec size: none | inline | full
- Examples: S-1, S-2
- Verification: existing or proposed test files

Feature:

## Goal
[What this feature does]

## Design
[UI/UX considerations, Figma links if mentioned]

## Acceptance Criteria
- [ ] ...

## Example Scenarios
- S-1: Given ..., when ..., then ...

## Spec Readiness
- Spec size: none | inline | full
- Full spec required before implementation: yes/no

Idea:

## Concept
[The idea]

## Why
[Why this could be valuable]

## Open Questions
- ...

Research:

## Question
[What we need to learn]

## Context
[Why this matters now]

## Deliverables
- [ ] Summary document / findings
- [ ] Recommendation

Rules

  • Always show the task before creating it — never create silently
  • Always include a description — even if the user provided a one-liner, expand it into at least 2-3 sentences
  • Always suggest labels — at minimum one type label (Bug/Feature/Improvement/Idea)
  • Detect duplicates — before creating, search existing open tasks for similar titles. If a match is found, show it and ask if this is a duplicate
  • Title max 80 chars — truncate and move detail to description if needed
  • No empty fields — every task must have: title, description, at least 1 label, priority
  • Idea detection — if the input is a question, starts with "maybe", "what if", "consider", "should we", add the Idea label automatically

Signals

GitHub stars
46
Forks
3
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
new-task
Source
github.com/glowingkitty/openmates