Skip to content

spec-check

A cross-repo spec engine. Durable domain requirements that survive past production — addressed by permanent ID, bound to code, and checkable in CI.

Keyed to the domain, not the feature

Feature-generation tools end at “implemented.” spec-check is keyed to the domain, which outlives every feature. Its first-class operation is life after production — superseding established truth, not generating new work.

Permanent IDs, mechanical integrity

Requirements carry permanent KEY-NNN IDs; code points back with // @spec KEY-NNN. From that one convention, an agent can be routed to exactly what a task touches, and spec check reports anything broken, orphaned, or unverified.

Cross-repo drift detection

The new thing a single-repo file scan can’t do: catch drift and propagation across version-pinned members in one local command — who migrated, who’s still pinned behind, what’s unverified.

Derived index that owns nothing

Canonical truth lives in git. The bun:sqlite index is rebuildable at any time and exists purely for speed and cross-repo queries — delete it and rebuild to an identical result.