Evidence — whole suite and repository gates (issue-203)

The same commands CI runs (hooks.prePush = lint, typecheck, unit-test; local == CI by reference/tooling.md). Captured after the last change to the branch.

make test

$ make test
uv run --project cli python -m pytest -q cli
........................................................................ [ 96%]
.....................................................................    [100%]
1796 passed, 1 skipped in 68.00s (0:01:08)

1796 passed against 1782 before this work item: the 6 precedence/abuse tests (2 of them parametrised over both transports, so 9 test ids), the 3 integration scenarios, and 2 regression guards. The single skip is pre-existing and unrelated.

make lint

$ make lint
uv run ruff check cli hooks
All checks passed!
npx --yes markdownlint-cli2@0.18.1 "**/*.md"
Linting: 556 file(s)
Summary: 0 error(s)

Markdown is linted too, which is the gate the spec chain, the decision record and the docs page all pass through.

make format-check

$ make format-check
uv run ruff format --check cli hooks
189 files already formatted

make typecheck

$ make typecheck
uv run pyright cli
0 errors, 0 warnings, 0 informations

Pyright is why _slack() in the unit tests narrows its resolve() result with an isinstance(provider, _SlackBase) assertion instead of reaching through the Integration protocol: the protocol deliberately exposes call and nothing else, and the narrowing doubles as the claim that both transports share one URL resolver.

make validate

$ make validate
uv run python scripts/validate_config.py
… 7 configs
VALID   .the-loop/cli-config.yaml
VALID   skills/the-loop/templates/cli-config.yaml

All 7 configs VALID, including both CLI configs — neither of which sets an inline url, which is the point: the property is optional and a config written before this change is unchanged and still valid.