sd search <query> — case-insensitive substring match over issue title and description, with the full filter set layered on top (--status, --type, --assignee, --label, --label-any, --unlabeled, --priority, --priority-max, --limit, --sort, --format).
No index, no ranking, no embeddings — a linear scan of a JSONL file, which is all the data model needs. Its role in the workflow is duplicate avoidance: the seeds-issue-workflow skill puts sd search "<keywords>" in the pre-flight step, before filing, so an agent claims existing work instead of forking a second seed for it.
A pure read command; maps to no canonical SDLC stage.
See Also
- seeds-list — filter-based rather than text-based listing.
- seeds-issue-workflow — the skill that mandates a search before a create.
- seeds — the parent framework.