Skip to content

Glossary

Canonical terms for AI Native SDLC engineering documentation.

Terms used across engineering docs. Prefer these definitions over ad-hoc re-explanations in each page.

TermDefinition
Runtime harness.sdlc/ plus materialization — operational infrastructure for governed cognitive execution
Runtime adapter{specialization_layer}/ — binding to a specific agent environment (Cursor, Claude, etc.)
MaterializationDeterministic projection of DSL declarations into verifiable files, gates, and manifests
Procedural DSL.sdlc/sdlc.yaml — executable operational specification
PlaybookAtomic domain procedure (.pb.yaml) — provisioned on demand, not loaded at session start
SkillSDLC orchestration composition (.cursor/skills/) — may declare playbook dependencies
GateDeterministic script verifying a stage transition; agent claims do not satisfy gates
Hydration scopeFixed INDEX sections loaded at session start; playbooks excluded until provision
WorktreeIsolated git checkout per Plane ticket — mandatory for feature/fix implementation
Cognitive traceRun manifest + trace UI — where the engineering process is visible
Runtime traceevents.jsonl — what the agent did (tools, hooks, spans)
target_rootApplication path bound by workspace config (e.g. app/)
BaselineVersioned SDLC pack snapshot — consumer vs authoring layers (ADR-014)
Learn pipeline/sdlc:learn — acquire domain capability as registered playbooks without contaminating agents

If a term is missing, add it here before introducing it in another doc.