Research: Plan Concern Vocabulary and WP Traceability
Pydantic v2 optional field with validator — pattern
Decision: Use Field(default_factory=list) with @field_validator("plan_concern_refs", mode="before") for the new list field.
Rationale: The existing wps_manifest.py already uses this exact pattern for dependencies, owned_files, requirement_refs, and subtasks. Consistency is the primary driver.
Alternatives considered: Optional[list[str]] = None — rejected because None vs [] distinction creates unnecessary null-checking downstream in generate_tasks_md_from_manifest().
IC-## pattern validation and ASCII constraint
Decision: Validate with re.match(r"^IC-\d{2}$", v) inside field_validator. Do not use \w or Unicode-aware patterns.
Rationale: Project directive DIR-010 requires ASCII-only identifiers using an explicit ASCII allowlist. IC-\d{2} covers IC-01 through IC-99 — sufficient for any realistic mission plan.
Alternatives considered: IC-\d+ (variable length) — rejected; fixed two-digit format matches WP## discipline and makes sorting predictable.
cross_cutting field placement
Decision: Add cross_cutting: bool = False to WorkPackageEntry alongside plan_concern_refs.
Rationale: The spec requires finalize-tasks to warn for WPs missing both plan_concern_refs and cross_cutting. Both fields live on WorkPackageEntry to keep the validation logic in one place. False default preserves backwards compatibility.
Alternatives considered: A sidecar YAML field outside WorkPackageEntry — rejected; the pydantic model is the schema authority and unknown fields in wps.yaml are either rejected or silently dropped depending on model config; a typed field is safer.
generate_tasks_md_from_manifest() rendering extension
Decision: Add a "Plan concerns: IC-01, IC-03" line to each WP block in tasks.md when plan_concern_refs is non-empty. Render nothing (no blank line, no label) when the list is empty.
Rationale: Matches existing rendering pattern — owned_files and requirement_refs are only rendered when non-empty. Keeps generated tasks.md clean for legacy missions.
Template section naming: "Implementation Concern Map"
Decision: Use ## Implementation Concern Map as the replacement section header.
Rationale: "Map" signals that the section is a structured enumeration of named units (not a prose description). "Implementation Concern" is unambiguous in the SDD context — it does not collide with "implementation" (the execute phase), "concern" (a general English word), or any existing spec-kitty term.
Alternatives considered: "Architecture Concerns", "Plan Decomposition", "Work Concerns" — all rejected as either too vague or too close to existing terminology.
Snapshot regeneration scope
Decision: Regenerate only tests/specify_cli/skills/__snapshots__/ snapshots. The twelve-agent regression baseline (tests/specify_cli/regression/_twelve_agent_baseline/) should be inspected for stale "Parallel Work Analysis" fixture content and updated if found — they are test fixtures, not historical artifacts.
Rationale: The renderer test suite (test_command_renderer.py) runs PYTEST_UPDATE_SNAPSHOTS=1 to regenerate. The twelve-agent baseline is a separate suite; update only entries that fail due to this change.