infer-research-from-code

What problem does it solve?

It turns scan-derived codebase evidence into per-domain product-scoped research documents, so teams can move forward with deterministic discovery and avoid blank or speculative research when there is no prior product knowledge.

Core Features & Use Cases

• Evidence-grounded research generation: Creates research/{domain}.md for each selected domain using scan-index.json evidence tied to that domain.
• Canonical KB extension with explicit gaps: Inherits the five KB-extension sections from the domain KB when available, and marks missing evidence with inline knowledge_gap indicators instead of fabricating content.
• Resolution trace and decision surfacing: Produces a decision manifest and a resolution trace so inferences and knowledge gaps are auditable and recoverable by later enrichment workflows.
• Aligned with codify codification flow: Runs strictly after domain-selection inference and emits artifacts into the STM evidence path for downstream promotion.

Quick Start

Run the skill during the codify stage for a given issue, providing scan-index.json plus the domain-selection.yaml, and it will output one research/{domain}.md file per selected domain into the codify evidence directory.