- Home
- Skills
- Yousufjoyian
- Claude Skills
- Task Create
task-create_skill
- Python
0
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill yousufjoyian/claude-skills --skill task-create- SKILL.md14.1 KB
Overview
This skill distills a conversation into a comprehensive task briefing and commits it to the active project's tasks/ directory. It produces a structured task JSON that gives another agent or a human everything needed to pick up the work cold. It shows the generated JSON for approval and updates the tasks registry on confirmation.
How this skill works
The skill scans the recent conversation to identify the problem, decisions, constraints, files mentioned, and any proposed approach. It builds a task object following the comprehensive format (id, title, context, why, what, done_when, etc.), prints the full JSON for user review, and on approval writes tasks/{nextId}/task.json and updates tasks/_registry.json. If no approach was discussed, it leaves optional fields out so the task stays open-ended.
When to use it
- You want a durable, handoff-ready task from a discussion instead of a quick note.
- A future agent or teammate must be able to complete work without re-reading the conversation.
- You hear backlog phrases like 'add that to the backlog', 'we should do that later', or 'queue this'.
- You need both human-readable context and machine-readable checklist/verification details.
- You want to preserve decisions and constraints raised during a live discussion.
Best practices
- Capture the minimal required fields always: id, title, context, why, what, done_when, status, priority, assignedProject, createdAt.
- Omit optional keys that do not apply rather than leaving empty values; include files as [] if unknown.
- Keep titles imperative and under 100 characters so scanning the backlog remains fast.
- Provide tangible verification steps and a small actionable checklist to enable live progress tracking.
- Show the JSON to the user for approval before writing files and updating the registry.
Example use cases
- Convert a design chat about per-tab session isolation into a task with context, risks, and checklist items.
- Turn 'let's come back to this' during a sprint meeting into a backlog task with priority and done_when criteria.
- Create a task for a bug observed during QA that includes reproduction steps, affected files, and verification steps.
- Capture a multi-phase feature discussion into a task with suggested strategy and phase-based checklist.
FAQ
The skill omits or leaves blank optional fields like how or strategy so the task remains an open problem for implementers to design.
How are checklist and status maintained?
Checklists are live; agents toggle items as they complete them. Task status is derived automatically from checklist progress (pending, in_progress, completed).