seeds-issue-workflow

name	seeds-issue-workflow
description	Turn a request into tracked, ready seeds work — file an issue with priority and labels, wire dependencies, and decompose large work into structured child seeds with sd plan. Use when an agent needs to create or organize work in the seeds repo.
tools	["sd","git"]
inputs	["a description of the work (a sentence or a multi-step feature)"]
outputs	["one or more seeds in .seeds/issues.jsonl, ready (unblocked) or correctly blocked","for large work, a plan row in .seeds/plans.jsonl with one child seed per step"]

seeds-issue-workflow

Use this skill inside the seeds repo when you need to turn a vague ask into a tracked unit of work that sd ready will surface. It encodes two procedures: create-then-ready for a single scoped task, and plan-decomposition for work large or ambiguous enough that an LLM benefits from structured breakdown first.

Pre-flight

sd prime                          # inject rules + command reference
sd ready                          # see what is already unblocked
sd search "<keywords>"            # avoid filing a duplicate

If a seed already covers the work, claim it (sd update <id> --status in_progress) instead of filing a new one.

Procedure A — create-then-ready (single scoped task)

For a task that fits in one focused commit:

sd create --title "Fix sd close race on concurrent writers" \
  --type bug --priority 1 \
  --description "Two agents closing the same id can drop one update."

Capture the printed id (e.g. seeds-1a2b), then tag and wire it:

sd label add seeds-1a2b concurrency
sd dep add seeds-1a2b seeds-9f8e   # this work depends on seeds-9f8e first

A seed is ready when it is open and every id in its blockedBy list is closed. Add dependencies for real ordering only — every unnecessary sd dep add hides the work from sd ready. Verify:

sd ready | grep seeds-1a2b        # appears once its blockers close

Procedure B — plan-decomposition (large / ambiguous work)

When the work is multi-step, emit a structured prompt, fill it, and let seeds spawn the children:

sd plan prompt seeds-1a2b > /tmp/plan.json   # emit the template to fill
$EDITOR /tmp/plan.json                       # write one step per unit of work
sd plan submit seeds-1a2b --plan /tmp/plan.json

Each step becomes one child seed; a step's blocks: [j] wires step i's id into step j's blockedBy, so the children come out in dependency order. Steps may carry labels: [...] that flow to the spawned child (normalized and merged additively). Inspect and track:

sd plan show <pl-id>              # sections, children, sub-plans
sd ready                          # the first unblocked child surfaces

Finish

Close completed work and sync before handing off:

sd close seeds-1a2b --reason "Fixed; covered by store.test.ts"
sd sync                           # stage + commit .seeds/ changes

Do not git push unless the user asks. Never hand-edit .seeds/issues.jsonl — always go through sd so advisory locks and atomic writes are honored.