infer-research-from-code

Generate domain research from code evidence by using scan-index.json and the canonical domain KB to produce per-domain Markdown files. This deterministic inference prevents speculative content by marking missing evidence with inline knowledge_gap indicators instead of fabricating data.

Deterministic inference for product codification is the process of deriving structured domain research from code evidence using explicit inclusion rules. It ensures auditable inference by emitting decision-manifest and resolution-trace artifacts alongside the generated Markdown files.

Document knowledge gaps in domain research files by using inline knowledge_gap markers where code evidence is absent. This approach ensures missing information is explicitly flagged for later enrichment workflows rather than filled with fabricated assumptions.

Use code evidence instead of web research for domain discovery during the codify stage when no web research and no prior product LTM are available. This makes the codebase the primary evidence surface for generating per-domain Markdown research.

Yes, the codify workflow requires a domain-selection.yaml file to generate research. The process runs strictly after domain-selection inference to produce scoped Markdown files and resolution-trace artifacts in the STM evidence path.

Deriving domain research solely from scan-index.json limits outputs to available code evidence, explicitly marking absent data as knowledge gaps. It cannot fabricate missing domain context and relies on downstream promotion workflows to resolve flagged gaps.