speckit-extract-run

To extract ADRs from spec directories, run speckit-extract-run to scan research.md, spec.md, data-model.md, contracts/, and plan.md. It categorizes decisions, rationales, and alternatives, then outputs ADRs directly into your dev/adr folder with source tagging.

The best way to automate knowledge and guidelines extraction across multiple specs is using batch extraction mode. It scans spec artifacts, categorizes content, and aggregates updates into dev/knowledge and dev/guidelines directories while maintaining an extraction ledger for full traceability.

Yes, you can process multi-spec batches simultaneously. The tool applies batch extraction logic across multiple completed spec directories, categorizing content into ADRs, knowledge, or guidelines, and produces a unified extraction ledger with source tagging.

Spec directory extraction categorizes content by scanning research.md, spec.md, data-model.md, contracts/, and plan.md. It identifies decisions, rationales, and alternatives for ADRs, and routes general knowledge or guidelines into their respective dev/ subdirectories.

You need completed spec directories containing research.md, spec.md, data-model.md, contracts/, and plan.md. The extraction process automatically generates the target structure, producing ADRs under dev/adr, knowledge under dev/knowledge, and guidelines under dev/guidelines.

Spec extraction maintains an extraction ledger with source tagging to improve traceability and onboarding. By linking extracted ADRs, knowledge, and guidelines back to their original spec artifacts, teams can verify decisions and track documentation origins.