spec-to-issueslisted
Install: claude install-skill luminik-io/alfred
# Spec to issues
## When to use
- A spec or roadmap item is APPROVED and someone asks to turn it into work.
- You are handed a design doc and told to "make issues out of this".
- A feature touches more than one repo and needs to be split so each issue is
single-repo and independently mergeable.
- Before any implementation firing runs: a good issue is what makes an
autonomous run land cleanly instead of sprawling.
Do NOT use this to invent scope. If the spec is a draft or the goal is unclear,
stop and say so. Issues derived from a fuzzy spec produce fuzzy PRs.
## Procedure
1. Read the spec. Confirm it has the minimal shape (goal, current behavior,
target behavior, acceptance criteria, out-of-scope). If it does not, see
`references/spec-shape.md` and ask for the missing pieces before scoping.
2. Identify the repos the spec touches. One issue per repo. If a single repo's
work is large, split it along a seam a reviewer can verify independently
(an endpoint, a screen, a migration), never along "part 1 / part 2" of the
same file.
3. Order the issues by dependency. If `your-backend` must ship an endpoint
before `your-frontend` can call it, say so in each issue's first line
("Depends on: your-backend #NN") so the fleet does not start the dependent
work early.
4. For each issue write:
- A one-line title in the form `feat(<area>): <verb> <thing>`.
- A short context paragraph: what the spec wants and why this slice exists.
- Acceptance criteria