Spec Kitty

Implementation Plan: Deep ROI pass over the large tests/charter suites

Branch: feat/6485-charter-tests-deep-roi-pass (stacked on PR #6483) | Date: 2026-10-19 | Spec: spec.md

Summary

Review the in-scope tests/charter/ files function by function against the issue #6485 rubric, and delete or consolidate tests that do not guard a live contract. Work is sliced into six work packages by file cluster, with disjoint file sets so they run as parallel lanes. Each work package writes its own ROI ledger. No product code changes.

Technical Context

Language/Version: Python 3.11+ (test code only) Primary Dependencies: pytest, pytest-xdist, pytest-cov (existing dev group; no dependency change) Storage: N/A Testing: pytest tests/charter tests/charter_offering -n auto --dist loadfile; coverage via --cov=src/charter Target Platform: Linux/macOS/Windows CI (unchanged) Project Type: single project — test-suite remediation Performance Goals: tests/charter/ wall-clock does not increase (NFR-003) Constraints: no src/ edits (C-001); no tests/architectural/ edits (C-002); ≤ 1.0 pp src/charter coverage drop (NFR-002) Scale/Scope: ≈ 20k lines across ~45 files in tests/charter/

Supply chain: no dependency is added, upgraded or removed, so DIRECTIVE_051 has no control to evaluate.

Charter Check

• Test remediation (Standing Order #4, DIRECTIVE_041): judge the test, not git-blame; stub/change-detector → delete. This mission applies it. PASS.
• Red-first / ATDD (C-011): no behaviour is added or fixed, so no failing-first test exists to write. The acceptance evidence is instead the ledger plus a green suite and a coverage delta. Recorded as a justified deviation in Complexity Tracking.
• No full heavy suites (NO_FULL_HEAVY_SUITES_IN_MISSION): WPs run only their own files plus the covering tests they name; the orchestrator runs tests/charter + tests/charter_offering for the before/after coverage. PASS.
• Tiered rigour / reviewer ≠ implementer: review by an opus reviewer-renata, implement by sonnet python-pedro. PASS.
• Terminology canon: no user-facing prose beyond the changelog. PASS.

Project Structure

Documentation (this mission)

kitty-specs/charter-tests-deep-roi-pass-01M5AN9J/
├── spec.md
├── plan.md
├── research.md
├── data-model.md
├── quickstart.md
├── roi-ledger/WP0N.md   # one per WP, written during implement
└── tasks.md / tasks/    # /spec-kitty.tasks

Source Code (repository root)

tests/charter/            # edited (deletions / consolidations only)
tests/charter_offering/   # read-only: covering tests
tests/architectural/      # read-only: covering tests

Complexity Tracking

Deviation	Why needed	Simpler alternative rejected because
No red-first test per WP	Pure test removal adds no behaviour to pin	A synthetic failing test would itself be a tautology this mission deletes

Implementation Concern Map

The baseline coverage and test count before the mission are recorded in research.md.

IC-01 — Context family

test_context.py, test_context_selection_render.py, test_context_render_seams.py, test_context_service_seams.py, test_context_leaf_seams.py, test_context_profile_lineage_parity.py; cluster 2 (activation/test_context_identity_reuse.py ↔ activation/test_context_org_root_identity.py).

IC-02 — Synthesizer

synthesizer/test_synthesize_reconcile.py, test_orchestrator_resynthesize.py, test_bundle_validate_extension.py, test_manifest.py; cluster 1 (test_synthesize_path_parity.py ↔ test_orchestrator_synthesize.py).

IC-03 — Packs and org

packs/test_pack_validator.py, activation/test_org_charter.py, activation/test_org_charter_merge_parity.py, test_org_drg_loader.py, test_pack_manager.py; cluster 6 (test_retired_pack_fields.py ↔ test_retired_accompanies_doctrine_pack.py).

IC-04 — Resolver, activation, compiler

test_resolver.py, test_resolver_activation_parity.py, test_cascade.py, test_compiler.py, test_mission_type_profiles.py, test_mission_type_activations_seed_read_parity.py; cluster 3 (activation/test_effective_set.py ↔ activation/test_effective_set_chain.py).

IC-05 — Remaining parity, schemas, cwd, sync

test_config_stem_parity.py, test_charter_generate_scoped_reference_parity.py, test_spdd_reasons_activation_parity.py, test_kind_vocabulary_recursion_parity.py, test_check_graph_kind_parity.py, test_direct_write_kinds_parity.py; clusters 4 (cwd), 5 (test_schemas*.py) and 7 (test_sync_paths.py ↔ test_sync.py).

IC-06 — Heavy tests (operator-added)

test_compiler_charter_yaml.py, test_context_unknown_language.py, test_regenerate_rederives_languages.py, test_progressive_disclosure.py, test_every_load_delivery.py, test_active_languages_idempotency.py, test_retrospective_governance_emit.py, test_activation_preserves_effective_4253.py. Same rubric, plus cost reduction that keeps the contract (FR-007).

Each IC owns a disjoint file set; a merge target named in a ledger must be inside the same IC's file set.