- Home
- Skills
- Arjenschwarz
- Agentic Coding
- Starwave Creating Spec
starwave-creating-spec_skill
- Python
16
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 arjenschwarz/agentic-coding --skill starwave-creating-spec- SKILL.md6.0 KB
Overview
This skill initializes a new feature spec and orchestrates the entire spec-driven workflow from a feature idea to an actionable task list. It guides you through sizing, requirements in EARS format, design, task breakdown, and optional branch creation with explicit approval gates. Transit ticket tracking is supported and kept in sync throughout the process.
How this skill works
I start with a scope assessment that inspects the codebase, estimates LOC and files affected, and recommends either a smolspec or a full spec workflow. If you approve, I run the required sub-steps: requirements gathering (EARS), design creation, task planning, and then offer branch creation. When a Transit ticket (T-#) is mentioned, I update its status at key checkpoints using mcp__transit__update_task_status and include required comments.
When to use it
- You're starting a new feature and need a documented, actionable spec and task plan.
- You want to ensure correct sizing before committing to a full spec or a smolspec shortcut.
- You need design decisions and code tasks captured for implementation and reviews.
- You must coordinate changes across multiple subsystems or teams.
- You have a Transit ticket to track through spec and implementation readiness.
Best practices
- Always confirm the scope assessment before proceeding to requirements or design.
- Record every decision immediately in decision_log.md and show it during reviews.
- Prefer full spec when uncertain or when changes touch multiple files or APIs.
- Keep tasks strictly coding-focused and include test tasks for TDD coverage.
- Offer ticket-based branch names first when a Transit ticket is present.
Example use cases
- A product manager proposes a new user dashboard widget that touches several services — run full spec to capture requirements and design.
- A small UI tweak affecting one component and one file — run smolspec to get a short spec and branch quickly.
- A backend API change that may be breaking — use full spec, document compatibility decisions, and plan migration tasks.
- Coordinating a cross-team feature that requires design decisions, task split, and a tracked Transit ticket.
- Preparing a ready-for-implementation task list and creating a feature branch with a ticket-aware naming convention.
FAQ
Use smolspec when the change is small: under ~80 LOC, 1–3 files, single component, no API breaks or cross-cutting concerns.
How are Transit tickets updated?
When a T-# is present I call mcp__transit__update_task_status to move it to spec after scope approval, and to ready-for-implementation after branch creation, always including the required comment.