Evidence — tests (issue-179)

Red first, then green. Every command run from this repository's checkout on 2026-08-08.

The red run — the new tests against the pre-change source

The four source files (pdlc-work-item-loop.yaml, hooks/artifacts.py, hooks/selection.py, templates/execution-log.md) were reverted to HEAD with the new tests in place:

$ uv run --directory cli pytest -q tests/test_graph_skips.py tests/test_graph_hooks.py \
    tests/test_graph_verification_integration.py tests/test_graph_parity.py \
    tests/test_core_graphs.py tests/test_graph_model.py
FAILED tests/test_graph_skips.py::test_shipped_outer_loop_marks_every_phase_but_the_gate_skippable
FAILED tests/test_graph_skips.py::test_shipped_skip_sets_name_the_two_chains
FAILED tests/test_graph_skips.py::test_the_former_floor_is_now_declarable_and_carries_no_required_marker
FAILED tests/test_graph_skips.py::test_the_checklist_says_so_when_nothing_is_protected
FAILED tests/test_graph_skips.py::test_the_shipped_checklist_offers_every_phase_the_item_walks
FAILED tests/test_graph_hooks.py::test_only_when_skipped_is_dormant_when_the_artifact_exists
FAILED tests/test_graph_hooks.py::test_only_when_skipped_is_dormant_when_nothing_was_declared
FAILED tests/test_graph_hooks.py::test_only_when_skipped_can_never_widen_what_may_be_skipped
FAILED tests/test_graph_verification_integration.py::test_verification_gates_the_log_when_the_plan_was_declared_away
FAILED tests/test_core_graphs.py::test_skip_declares_against_the_shipped_vocabulary
FAILED tests/test_graph_model.py::test_the_shipped_graph_compiles - AssertionError
11 failed, 142 passed in 0.90s

Two notes on what did not go red, because a green test in a red run is worth explaining:

• test_only_when_skipped_gates_the_fallback_for_a_planned_absence (M7) passes either way — without the parameter the entry simply applies unconditionally, which produces the same block-then-pass. Its three siblings (M8) are what pin the dormancy, and all three were red.
• test_declaring_every_phase_away_walks_the_item_to_its_terminal (M6) runs against a fixture graph, not the shipped one, so it does not depend on the YAML edit.

M13 — the parity red, isolated

test_p5c only fails once the graph gates the new section, so it was run with the source restored and only templates/execution-log.md at HEAD:

$ uv run --directory cli pytest -q tests/test_graph_parity.py
E  AssertionError: a node gates a section of a shared artifact that the bundled template
   does not offer, so every work item authored from the template blocks there:
   execution-log.md: node 'verification' requires a 'Verification results' section the
   template does not offer
1 failed, 7 passed in 0.20s

This is issue-167's failure mode caught by the test issue-167 produced — exactly the regression the section addition prevents.

The green run

Each file run on its own (uv run --directory cli pytest -q <file>):

File	Result
tests/test_graph_skips.py	46 passed
tests/test_graph_hooks.py	46 passed
tests/test_graph_verification_integration.py	10 passed
tests/test_graph_parity.py	8 passed
tests/test_core_graphs.py	5 passed

Full regression

$ uv run --directory cli pytest -q
1480 passed, 1 skipped in 47.32s

Baseline on main before this work item: 1467 passed, 1 skipped. The delta is +13: twelve new tests (M1–M13 minus the ones folded into existing cases) and one existing test replaced by two (test_shipped_outer_loop_marks_exactly_the_spec_chain_skippable and test_shipped_floor_is_not_skippable became four assertions of the new truth).

The shipped graph, as compiled

$ uv run --directory cli python -c "..."
skippable:     brainstorming, capability-docs, critic-review, design, design-approval,
               evidence, human-approval, implementation, requirements-approval,
               requirements-definition, reviewer-briefing, security-review, self-review,
               tasks-breakdown, test-planning, verification
not skippable: complete, escalated, phase-selection
required:      phase-selection
sets:          spec-chain  = brainstorming, requirements-definition,
                             requirements-approval, design, test-planning,
                             design-approval, tasks-breakdown
               review-chain = self-review, critic-review, security-review, evidence,
                             capability-docs, reviewer-briefing