kiro-discovery

Scoping a vague feature request involves performing a lightweight metadata scan of existing specs and repository structure to determine the correct action path. This guides sequential clarification questions to elicit stakeholder intent and boundaries, recommending whether to create a new spec, update one, or implement directly.

Early project intake and discovery requires evaluating existing spec coverage, steering documents, and repository metadata. A structured discovery process proposes 2-3 approaches with trade-offs and produces actionable artifacts like brief.md or roadmap.md to define boundaries and next steps.

Decomposing a product change request requires targeted clarification questions to map work boundaries. The process inventories existing specs and steering files, then generates a brief.md or roadmap.md artifact while recommending the exact next CLI-like command for continuing the workflow.

You need a new spec when a lightweight metadata scan reveals no existing spec coverage for the requested changes. Direct implementation is recommended when existing steering documents and repository metadata already provide clear boundaries and ownership for the bug report.

Structured discovery handles multi-scope enhancement requests by proposing 2-3 approaches with defined trade-offs. It produces roadmap.md artifacts to capture the multiple scopes and recommends the exact next CLI-like command for continuing the decomposition process.

Lightweight metadata scans for requirements discovery intentionally avoid loading heavy file contents to quickly inventory specs and repo structure. The limitation is that deep contextual logic within existing files is not evaluated, meaning highly complex or deeply nested code changes may require further manual investigation beyond the initial scan.