Spec Kitty

└─ kitty-specs
   └─ Doctrine Tension as First-Class DRG Edges

Mission Run:

📚 Docs ↗

description: "Work package task list for Doctrine Tension as First-Class DRG Edges"

Work Packages: Doctrine Tension as First-Class DRG Edges

Inputs: kitty-specs/doctrine-tension-edges-01KY1WPC/{spec.md, plan.md, research.md, data-model.md, contracts/, quickstart.md, occurrence_map.yaml} Prerequisites: plan.md, spec.md, data-model.md, contracts/, occurrence_map.yaml (this mission runs under change_mode: bulk_edit — every WP's diff is checked against occurrence_map.yaml at review time)

Tests: Included — spec.md's Correctness Invariants (INV-001..005) and Success Criteria (SC-001..005) are only provable with tests, and NFR-005 requires focused tests for every new branch/helper.

Organization: 41 subtasks (T001-T041) roll up into 8 work packages (WP01-WP08), sequenced per plan.md's Implementation Concern Map (IC-01..IC-07) and Constraint C-006's green-at-every-boundary migration order.

Prompt Files: Each work package references a matching file in tasks/. This file is the high-level checklist; implementation detail lives in the prompt files.

Path Conventions

Single project — src/doctrine/, src/charter/, src/specify_cli/, tests/, docs/. No frontend/backend split (per plan.md Project Structure).

Work Package WP01: Relation vocabulary + NodeKind foundation (Priority: P0)

Goal: Add the three Relation enum members, the relation-description registry (with final description text), the anti_pattern NodeKind, and DRGNode.tags. Independent Test: Relation.IN_TENSION_WITH/RECONCILES_TENSION/REJECTS and NodeKind.ANTI_PATTERN exist and round-trip through DRGNode/DRGEdge validation; RELATION_DESCRIPTIONS has all 3 entries. Prompt: tasks/WP01-relation-vocabulary-foundation.md Requirement Refs: FR-001, FR-002, FR-003, FR-004, FR-012

Included Subtasks

✅ T001 Add IN_TENSION_WITH, RECONCILES_TENSION, REJECTS to Relation (src/doctrine/drg/models.py)
✅ T002 Add RELATION_DESCRIPTIONS registry with final description text for the 3 new relations (src/doctrine/drg/models.py)
✅ T003 [P] Add ANTI_PATTERN to NodeKind + URN-prefix wiring in DRGNode._validate_urn (src/doctrine/drg/models.py)
✅ T004 [P] Add tags: list[str] = Field(default_factory=list) to DRGNode (src/doctrine/drg/models.py)
✅ T005 Add corresponding ArtifactKind member (src/doctrine/artifact_kinds.py)
✅ T006 Unit tests: enum values, RELATION_DESCRIPTIONS completeness, DRGNode.tags round-trip, anti_pattern:<id> URN acceptance (new tests/doctrine/drg/test_models.py; extend tests/doctrine/drg/test_nodekind_artifactkind.py and test_kind_mapping_totality.py)

Implementation Notes

T001/T002 are the same file, sequential. T003/T004 touch the same file but are independent edits — parallel-safe in intent, sequential in practice within one WP.
T006 extends two existing totality/mapping tests, not just the new file — test_kind_mapping_totality.py likely asserts every NodeKind maps somewhere; adding ANTI_PATTERN without updating its expectations will break it (a useful regression signal, not a bug to route around).

Parallel Opportunities

T003/T004/T005 touch different declarations and can be drafted concurrently by the same implementer before running tests.

Dependencies

None — foundation package, must land first (C-006 step 1).

Risks & Mitigations

ANTI_PATTERN ripples into _SINGULAR_TO_PLURAL/_SINGULAR_TO_PER_KIND_FIELD (WP04) and _SINGULAR_TO_PLURAL_KIND (WP04) — this WP only adds the enum member; WP04 owns the activation-filter wiring. Do not add those mappings here — it would create an ownership conflict.

Work Package WP02: Hand-author tension/reconciliation/rejection edges (Priority: P0)

Goal: Author the new-relation edges and marked anti-pattern nodes directly in the per-kind DRG graph fragments (*.graph.yaml) — the extractor has no frontmatter mechanism for these three relations, so they are hand-authored, not derived. Independent Test: Loading directive.graph.yaml/tactic.graph.yaml/paradigm.graph.yaml produces a graph containing the 024↔025 and smallest-viable-diff↔025 in_tension_with edges, the 6 anti-pattern nodes, the 8 rejects edges, and the new reconcile-change-scope-tensions directive with its 3 reconciles_tension edges. Prompt: tasks/WP02-hand-author-tension-edges.md Requirement Refs: FR-006, FR-007, FR-011

Included Subtasks

✅ T007 Hand-author in_tension_with edge, directive:024-locality-of-change ↔ directive:025-boy-scout-rule (lex-smaller URN as source) in src/doctrine/directive.graph.yaml
✅ T008 Hand-author in_tension_with edge, tactic:change-apply-smallest-viable-diff ↔ directive:025-boy-scout-rule (INV-005 — the recovered tactic↔directive tension) across tactic.graph.yaml/directive.graph.yaml
✅ T009 Create the 6 marked anti-pattern/smell nodes (anemic-domain-model, big-ball-of-mud, big-upfront-design, code-is-the-documentation, database-driven-design, single-diagram-architecture) with kind: anti_pattern + tags in paradigm.graph.yaml
✅ T010 Author the 8 rejects edges from the relevant paradigms to the 6 nodes from T009 in paradigm.graph.yaml
✅ T011 Create src/doctrine/directives/built-in/reconcile-change-scope-tensions.directive.yaml (new built-in directive; body explains how to weigh the 024/025/smallest-viable-diff tensions) + its reconciles_tension edges to all three in directive.graph.yaml/tactic.graph.yaml
✅ T012 Verify the new content loads without validation errors (existing graph-load tests) and record how the shipped-graph freshness canary will need to tolerate these hand-authored edges once WP03 regenerates the graph (do not modify the canary here — that is WP03's job)

Implementation Notes

This is authored content, not a schema/model change — depends on WP01 for the enum members to exist (Pydantic will reject an edge referencing an unknown Relation).
T011's new directive needs ordinary requires/scope edges too if applicable — those remain extractor-derived from its frontmatter as normal; only the reconciles_tension edges are hand-authored.
Do not remove any opposed_by content here — that is WP03, sequenced after this WP so the pack never loses tension information mid-migration (C-006).

Parallel Opportunities

T007/T008 (directive/tactic graph) and T009/T010 (paradigm graph) touch different files and can proceed in parallel.

Dependencies

Depends on WP01.

Risks & Mitigations

The shipped-graph freshness canary (test_shipped_graph_yaml_is_fresh et al.) currently expects the committed graph to be byte-identical to a fresh extractor run — WP03 must reconcile that expectation with these hand-authored additions surviving regeneration. Flag this explicitly in the WP02→WP03 handoff rather than silently hoping WP03 notices.

Work Package WP03: Retire `opposed_by` / `Contradiction` (Priority: P0)

Goal: Remove the opposed_by field/schema-property, the extractor logic that mints replaces from it, and the Contradiction model — in the C-006-mandated order (YAML+extractor together, then schema, then field+model together) — without losing WP02's hand-authored edges. Independent Test: grep -rn "opposed_by" src/ docs/ tests/ returns zero hits; doctrine.shared.models.Contradiction and its __all__ entry are gone; the dead-symbol gate (scoped to that symbol) is green; the 024↔025 replaces cycle no longer exists in the regenerated graph and WP02's in_tension_with/rejects/reconciles_tension edges survive. Prompt: tasks/WP03-retire-opposed-by.md Requirement Refs: FR-005, NFR-002

Included Subtasks

✅ T013 Remove the opposed_by-minting block(s) from src/doctrine/drg/migration/extractor.py (the block that emits a Relation.REPLACES edge from an opposed_by frontmatter entry)
✅ T014 Remove opposed_by: from the 5 built-in YAML sources (024-locality-of-change.directive.yaml, 025-boy-scout-rule.directive.yaml, change-apply-smallest-viable-diff.tactic.yaml, and the 3 paradigm files: brownfield-onboarding, c4-incremental-detail-modeling, domain-driven-design)
✅ T015 Remove the opposed_by property + contradiction definition from src/doctrine/schemas/{directive,paradigm,tactic}.schema.yaml
✅ T016 Remove the opposed_by field from src/doctrine/{directives,tactics,paradigms}/models.py; remove the Contradiction model + its __all__ entry from src/doctrine/shared/models.py; remove now-dead imports
✅ T017 Remove/repoint the opposed_by-specific tests: the extractor test at tests/doctrine/drg/migration/test_extractor.py (test_directive_opposed_by_produces_replaces), the 3 tests in tests/doctrine/test_directive_consistency.py, and update tests/doctrine/fixtures/paradigm/valid/with-tactic-refs.yaml
✅ T018 Regenerate the shipped *.graph.yaml fragments and reconcile the freshness canary so it accepts WP02's hand-authored edges while still catching genuine drift; confirm zero opposed_by hits and a green dead-symbol gate

Implementation Notes

Follow C-006's order strictly: T013+T014 together (blocks that stop minting AND the YAML that would otherwise still trigger them), then T015, then T016. Authoring the edge before its enum member (already done in WP01/WP02) or dropping the schema property before the YAML would produce a red state.
Anticipated small out-of-map edit: T018 touches src/doctrine/*.graph.yaml — files WP02 owns — only to regenerate/drop the now-stale opposed_by-derived replaces edges. Do not re-author or remove WP02's new edges; this is a one-line-rationale out-of-map touch (per the occurrence-map exception process), not a claim of ownership over those files.
The dead-symbol gate must be scoped to the Contradiction symbol specifically (module/class, not the bare word) — src/specify_cli/charter_runtime/lint/checks/contradiction.py's unrelated ContradictionChecker and its importers/tests are explicitly out of scope (see occurrence_map.yaml exceptions) and must not be touched.

Parallel Opportunities

T014/T015 (different files, same removal) can proceed in parallel; T016 should follow both (needs the schema gone first per C-006 step ordering intent, though Python import removal doesn't strictly require it — keep the stated order for reviewability).

Dependencies

Depends on WP02 (edges must exist before the field they replace is dropped).

Risks & Mitigations

Highest-risk WP for silent breakage (touches schema + models + extractor + built-in content + tests in one pass). Run the full tests/doctrine/ suite plus ruff/mypy before marking done, not just the touched test files.
This WP is the primary surface governed by occurrence_map.yaml's bulk_edit classification — review will run the diff-compliance check against it.

Work Package WP04: Read-surface wiring — validator + activation filter (Priority: P1)

Goal: Validate rejects targets (INV-004), wire the anti_pattern kind through the activation-filter's kind/per-ID gates, and confirm symmetric in_tension_with query (INV-001). Independent Test: A rejects edge to an unmarked node raises a validation error; an anti_pattern node's activation is gated the same way every other kind is gated; querying a stored in_tension_with edge from either endpoint returns the pair. Prompt: tasks/WP04-read-surface-wiring.md Requirement Refs: FR-004

Included Subtasks

✅ T019 Add "anti_pattern": "anti_patterns" to _SINGULAR_TO_PLURAL and the corresponding "activated_anti_patterns" entry to _SINGULAR_TO_PER_KIND_FIELD (src/charter/drg.py); mirror the plural mapping in _SINGULAR_TO_PLURAL_KIND (src/charter/activations.py)
✅ T020 Add activated_anti_patterns: frozenset[str] | None field to PackContext (src/charter/pack_context.py), following the existing per-kind field pattern
✅ T021 Add rejects-target validation to src/doctrine/drg/validator.py: a rejects edge whose target lacks kind == NodeKind.ANTI_PATTERN is a validation error (INV-004)
✅ T022 Test: symmetric read (INV-001) — a single stored in_tension_with edge is discoverable from both endpoint URNs via the graph's existing query helpers (new tests/doctrine/drg/test_validator.py or extend an existing DRG test)
✅ T023 Tests: anti_pattern kind activation-gating behaves like every other kind (extend tests/charter/test_drg_filtering.py / test_activation_filtered_drg.py); rejects-target validation error case (new tests/doctrine/drg/test_validator.py)

Implementation Notes

Verified, no code change needed: src/charter/cascade.py::_kind_of resolves generically via ArtifactKind(prefix) — once WP01's ArtifactKind member exists, _kind_of handles anti_pattern URNs correctly with zero edits. Do not touch cascade.py in this WP.
Before T019/T020, confirm whether anti_pattern nodes should default-allow (skip the per-kind config gate entirely, since they're validation targets rather than user-activatable content) or be config-gated like every other kind. FR-004 explicitly calls for wiring through _SINGULAR_TO_PLURAL/_SINGULAR_TO_PER_KIND_FIELD, so the spec's answer is config-gated — implement it that way even though _node_is_activated's default-allow-for-unknown-kind behavior would otherwise make this optional.

Parallel Opportunities

T019/T020 (activation-filter kind wiring) and T021 (validator) touch unrelated files and can proceed in parallel.

Dependencies

Depends on WP01 only (needs ANTI_PATTERN/ArtifactKind to exist). This WP's tests use synthetic constructed graphs, not the real built-in pack, so it needs neither WP02's authored edges nor WP03's removal — it runs in parallel with both.

Risks & Mitigations

Three near-duplicate singular→plural dicts (drg.py, activations.py, and potentially a third) must all move together or activation/CLI parsing diverges — cross-check all call sites of each dict before considering this WP done.

Work Package WP05: Consistency-check finding + activate warning (Priority: P1)

Goal: Advisory, always-on tension_unreconciled finding on ConsistencyReport (excluded from coherent, fail-closed on load errors) and the matching charter activate warning. Independent Test: Per contracts/tension-finding.md — co-activated unreconciled pair produces exactly one finding with both resolution paths; only-one-side-active produces none; removing the built-in reconciler makes the 024/025 finding reappear, restoring it clears it (SC-002). Prompt: tasks/WP05-consistency-check-and-activate-warning.md Requirement Refs: FR-009, FR-010

Included Subtasks

✅ T024 Add TensionFinding (sorted URN pair + the two fixed resolution-path strings) and unreconciled_tensions: list[TensionFinding] on ConsistencyReport (src/charter/consistency_check.py), excluded from the coherent reduction, and update to_json()
✅ T025 Implement the tension scan over the activation-filtered graph: one finding per co-activated in_tension_with pair, keyed on the sorted URN pair for dedup (Edge Case: symmetric authoring drift), with no transitive closure (INV-002 — A⋈B + B⋈C never synthesizes A⋈C)
✅ T026 Implement the reconciliation check: a pair is resolved only when an active artefact has reconciles_tension edges to both sides (half-reconciled — only one edge — does not resolve, US2 sc2)
✅ T027 Make the DRG load for this check fail closed into verification_errors — any exception during the scan surfaces there, never silently reduces to an empty finding list
✅ T028 Add the charter activate warning using the same TensionFinding shape, alongside existing activation warnings (locate the activate command in src/specify_cli/cli/commands/charter/)
✅ T029 Tests (contracts/tension-finding.md is the spec): NFR-001 positive assertion (finding present when warranted — a no-op returning [] must fail this test); non-finding case (only one side active); SC-002 before/after (remove reconcile-change-scope-tensions → findings appear; restore → clear); half-reconciled case (new tests/charter/test_tension_unreconciled.py; extend tests/charter/test_consistency_check.py)

Implementation Notes

T024-T027 are one logical unit (the finding + its computation) — do not split the dedup/non-transitivity logic from the fail-closed error handling; a scan that computes findings correctly but swallows exceptions fails FR-009 just as badly as one with no logic at all.
T028 depends on T024's TensionFinding shape existing first.

Parallel Opportunities

Limited — T024→T025→T026→T027 are a dependency chain within this WP. T028 can start once T024 lands.

Dependencies

Depends on WP01, WP02 (needs the built-in reconciler + tension edges to exist for SC-002's live assertion). No technical dependency on WP04 — this WP's scan only ever looks at in_tension_with/reconciles_tension edges among directive/tactic nodes, never rejects edges or anti_pattern nodes, so WP04's activation-filter wiring is orthogonal — this WP runs in parallel with WP04.

Risks & Mitigations

NFR-001's "a no-op checker returning [] fails this requirement" is a real trap — write the positive-finding test before the implementation (red-first) so it's impossible to accidentally ship a checker that never fires.

Work Package WP06: Orphan-lint fix + cascade exclusion regression (Priority: P1)

Goal: Remove the phantom governs/supersedes orphan-lint branches (FR-008, closes #2737) and add the regression test proving the three new relations stay out of cascade's REFERENCE_RELATIONS allowlist (FR-013). Independent Test: charter lint's orphaned_directive findings equal exactly {DIRECTIVE_035, DIRECTIVE_039}; activating one side of a tension never auto-activates the other (INV-003). Prompt: tasks/WP06-orphan-lint-and-cascade-exclusion.md Requirement Refs: FR-008, FR-013

Included Subtasks

✅ T030 Remove the governs (directive) and supersedes (adr) branches — and their module-docstring references — from _ORPHAN_RULES in src/specify_cli/charter_runtime/lint/checks/orphan.py (neither is a real Relation enum member)
✅ T031 Test: orphaned_directive findings equal exactly {DIRECTIVE_035, DIRECTIVE_039} — count == 2, zero false positives on referenced directives (extend tests/specify_cli/charter_lint/checks/test_orphan.py)
✅ T032 Add a regression test asserting {Relation.IN_TENSION_WITH, Relation.RECONCILES_TENSION, Relation.REJECTS} & cascade.REFERENCE_RELATIONS == frozenset() (extend tests/charter/test_cascade.py) — exclusion is by omission, so this test IS the FR-013 deliverable, not a supporting check for a code change
✅ T033 Regression test (INV-003): activating one side of a tension pair does not auto-activate the other; activating reconcile-change-scope-tensions does not activate 024/025/the tactic (new tests/charter/test_tension_cascade_exclusion.py)

Implementation Notes

T030 is the only source change in this WP — T032 requires no cascade.py edit (the three relations are simply never added to the allowlist); do not "fix" this by writing a denylist check, which would reintroduce the per-kind branching the cascade engine's design deliberately avoids (C-003).
Do not touch src/charter/cascade.py itself in this WP — WP04 already confirmed _kind_of needs no change, and REFERENCE_RELATIONS needs no change either (correct-by-omission).

Parallel Opportunities

T030/T031 (orphan-lint) and T032/T033 (cascade) are independent pairs and can proceed in parallel.

Dependencies

Depends on WP01 (needs the 3 relations to exist to assert their absence from the allowlist) and WP02 (INV-003's test needs the actual tension/reconciler edges to exist).

Risks & Mitigations

A cascade test that only checks "no crash" instead of "specific relations absent from the allowlist" would pass vacuously — assert the frozenset intersection explicitly, not just call-and-observe.

Work Package WP07: `spec-kitty migrate` downstream compatibility subcommand (Priority: P2)

Goal: New CLI subcommand rewriting org-pack-authored opposed_by → in_tension_with/rejects edges (FR-015, resolved decision: no deprecation window). Independent Test: Per contracts/migrate-opposed-by.md — --dry-run reports planned rewrites without writing; a real run rewrites and removes opposed_by; a second run against an already-migrated pack is a no-op; an unclassifiable entry produces a clear diagnostic, not a raw traceback. Prompt: tasks/WP07-migrate-opposed-by-subcommand.md Requirement Refs: FR-015

Included Subtasks

✅ T034 Implement the rewrite logic in a new src/specify_cli/migration/rewrite_opposed_by.py: classify each opposed_by entry (tension-style → in_tension_with edge; anti-pattern-rejection-style → rejects edge, creating/linking an anti_pattern-marked target node), modeled on migration/backfill_identity.py
✅ T035 Wire spec-kitty migrate rewrite-opposed-by [--pack PATH] [--dry-run] [--json] in src/specify_cli/cli/commands/migrate_cmd.py, following the existing subcommand conventions (e.g. backfill-identity)
✅ T036 Diagnostic path: an unclassifiable opposed_by entry exits non-zero with a clear message naming the entry and why it couldn't be classified — never a raw Pydantic validation traceback
✅ T037 Tests: idempotency (second run on an already-migrated pack is a no-op), --dry-run writes nothing, successful rewrite removes the opposed_by key, unclassifiable-entry diagnostic (new tests/specify_cli/migration/test_rewrite_opposed_by.py)

Implementation Notes

This WP is fully additive/new code — it does not touch any file another WP owns. It can start as soon as WP01's relations exist; it does not need WP02/WP03 to be done first (it operates on external org-pack YAML, not this repo's built-in content).

Parallel Opportunities

Can run fully in parallel with WP02-WP06 once WP01 is done.

Dependencies

Depends on WP01 only.

Risks & Mitigations

Without this command, removing opposed_by from additionalProperties: false schemas (WP03) breaks downstream consumers with only a validation error pointing at the symptom. Land this WP before or alongside WP03's release, not long after.

Work Package WP08: Relation doc-parity (Priority: P3)

Goal: Mirror the three new relations' descriptions into docs/architecture/doctrine-relationships.md and build the enum↔doc parity check (does not exist today — a deliverable per Assumption A2). Independent Test: Per US5 — parity check passes when registry and doc match; mutating one relation's description in the doc only makes the check fail, naming that relation (red-first). Prompt: tasks/WP08-relation-doc-parity.md Requirement Refs: FR-012

Included Subtasks

✅ T038 Add matching entries for the 3 new relations to docs/architecture/doctrine-relationships.md, verbatim against WP01's RELATION_DESCRIPTIONS text
✅ T039 Build the enum↔doc parity check comparing RELATION_DESCRIPTIONS values to the doc file's text for the 3 relations (new module/test — no such check exists today)
✅ T040 Red-first test: mutate one relation's description in the doc only → parity check fails naming that relation; revert → passes (new tests/doctrine/test_relation_doc_parity.py)
✅ T041 Wire the parity check into the test suite (or charter lint, if that is the more consistent home) so it runs automatically, not as a manual step

Implementation Notes

Read-only with respect to src/doctrine/drg/models.py — this WP consumes WP01's registry, it does not edit it. If the registry's placeholder text needs wordsmithing, that edit belongs in WP01, not here (avoids an ownership conflict on models.py).

Parallel Opportunities

Can run fully in parallel with WP02-WP07 once WP01 is done.

Dependencies

Depends on WP01 only.

Risks & Mitigations

A parity check that only verifies presence (both sides have a description) rather than content-equality would satisfy the letter and miss the point of NFR-004 — the red-first test in T040 is the guard against that shortcut.

Dependency & Execution Summary

Sequence: WP01 → WP02 → {WP03, WP04, WP06 in parallel (WP04 needs only WP01; WP03 and WP06 need WP02)} → WP05 (needs WP01+WP02 only, but is easiest to land once WP03 is settled to avoid reviewing against a moving target) → {WP07, WP08 in parallel from WP01}.
Parallelization: WP07 and WP08 need only WP01 and can run the entire time WP02-WP06 are in progress. WP04 needs only WP01 and can run in parallel with WP02/WP03. WP05 needs only WP01+WP02 and can run in parallel with WP03/WP04/WP06. Within WP02, directive/tactic-graph edits (T007/T008) and paradigm-graph edits (T009/T010) are independent.
MVP Scope: WP01 + WP02 + WP03 deliver the core data-model migration (US3 — opposed_by retired, tensions modeled as edges). WP05 is required to make the tension visible to an operator (US1 — the mission's actual reason to exist), so a genuinely usable MVP is WP01-WP05.

Requirements Coverage Summary

Requirement ID	Covered By Work Package(s)
FR-001	WP01
FR-002	WP01, WP05
FR-003	WP01
FR-004	WP01, WP04
FR-005	WP03
FR-006	WP02
FR-007	WP02
FR-008	WP06
FR-009	WP05
FR-010	WP05
FR-011	WP02
FR-012	WP01, WP08
FR-013	WP06
FR-014	WP01, WP02, WP03, WP04, WP05 (cross-cutting — see plan.md Change Surface Map)
FR-015	WP07
NFR-001	WP05
NFR-002	WP03
NFR-003	WP06
NFR-004	WP08
NFR-005	All WPs (cross-cutting quality gate — ruff/mypy/tests, not mapped to one WP by design)

Subtask Index (Reference)

Subtask ID	Summary	Work Package	Priority	Parallel?
T001	Add 3 new Relation enum members	WP01	P0	No
T002	Add RELATION_DESCRIPTIONS registry	WP01	P0	No
T003	Add ANTI_PATTERN NodeKind	WP01	P0	Yes
T004	Add DRGNode.tags field	WP01	P0	Yes
T005	Add ArtifactKind member	WP01	P0	Yes
T006	Foundation unit tests	WP01	P0	No
T007	Author 024↔025 in_tension_with edge	WP02	P0	Yes
T008	Author tactic↔025 in_tension_with edge (INV-005)	WP02	P0	Yes
T009	Create 6 anti-pattern nodes	WP02	P0	Yes
T010	Author 8 rejects edges	WP02	P0	Yes
T011	Create reconcile-change-scope-tensions directive	WP02	P0	No
T012	Verify new content loads; note freshness-canary handoff	WP02	P0	No
T013	Remove opposed_by-minting extractor blocks	WP03	P0	No
T014	Remove opposed_by from 5 built-in YAML files	WP03	P0	Yes
T015	Remove opposed_by/contradiction from 3 schema files	WP03	P0	Yes
T016	Remove opposed_by field + Contradiction model	WP03	P0	No
T017	Remove/repoint opposed_by-specific tests	WP03	P0	No
T018	Regenerate graph.yaml; reconcile freshness canary	WP03	P0	No
T019	Wire anti_pattern into activation-filter kind maps	WP04	P1	Yes
T020	Add activated_anti_patterns to PackContext	WP04	P1	Yes
T021	Add rejects-target validation (INV-004)	WP04	P1	Yes
T022	Symmetric-read test (INV-001)	WP04	P1	No
T023	Activation-gating + validation tests	WP04	P1	No
T024	Add TensionFinding + unreconciled_tensions	WP05	P1	No
T025	Implement tension scan (dedup, non-transitive)	WP05	P1	No
T026	Implement reconciliation check	WP05	P1	No
T027	Fail-closed DRG load error handling	WP05	P1	No
T028	Add charter activate warning	WP05	P1	No
T029	Consistency-check + activate tests	WP05	P1	No
T030	Remove phantom orphan-lint branches	WP06	P1	Yes
T031	Orphan-lint exact-set test	WP06	P1	Yes
T032	Cascade REFERENCE_RELATIONS exclusion test	WP06	P1	Yes
T033	Cascade non-auto-activation test (INV-003)	WP06	P1	Yes
T034	Implement rewrite_opposed_by module	WP07	P2	No
T035	Wire migrate rewrite-opposed-by CLI command	WP07	P2	No
T036	Unclassifiable-entry diagnostic	WP07	P2	No
T037	Migration idempotency + CLI tests	WP07	P2	No
T038	Add relation descriptions to doc	WP08	P3	Yes
T039	Build enum↔doc parity check	WP08	P3	No
T040	Red-first parity test	WP08	P3	No
T041	Wire parity check into test suite	WP08	P3	No

> 8 work packages, 41 subtasks. Average ~5 subtasks/WP — within the 3-7 ideal range on every WP.

Spec Kitty

Work Packages: Doctrine Tension as First-Class DRG Edges

Path Conventions

Work Package WP01: Relation vocabulary + NodeKind foundation (Priority: P0)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP02: Hand-author tension/reconciliation/rejection edges (Priority: P0)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP03: Retire opposed_by / Contradiction (Priority: P0)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP04: Read-surface wiring — validator + activation filter (Priority: P1)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP05: Consistency-check finding + activate warning (Priority: P1)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP06: Orphan-lint fix + cascade exclusion regression (Priority: P1)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP07: spec-kitty migrate downstream compatibility subcommand (Priority: P2)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Work Package WP08: Relation doc-parity (Priority: P3)

Included Subtasks

Implementation Notes

Parallel Opportunities

Dependencies

Risks & Mitigations

Dependency & Execution Summary

Requirements Coverage Summary

Subtask Index (Reference)

Work Package WP03: Retire `opposed_by` / `Contradiction` (Priority: P0)

Work Package WP07: `spec-kitty migrate` downstream compatibility subcommand (Priority: P2)