Verification (issue-374)

Executed at the verification node on 2026-09-30, head HEAD of PR #436, in the work item's cloud checkout (uv 0.12.21, Python 3.11, bun 1.x). One section per activity of testing-plan.md; raw summaries as the runner printed them.

T1 — unit

$ uv run --project cli python -m pytest -q cli/tests/test_instance_role.py cli/tests/test_instance.py \
    cli/tests/test_manager_fleet.py cli/tests/test_manager_stream.py cli/tests/test_instances_core.py \
    cli/tests/test_instances_cmd.py cli/tests/test_lifecycle_cmd.py cli/tests/test_api_config_integration.py
164 passed in 11.34s

Re-run after self-review round 3 (0b2cd86), which added six regression tests here. Red → green: test_instance_role.py was written first and failed on import (ImportError: MANAGER); test_manager_fleet.py went red on the fan-out's unbounded wait and the probe's detail text before the implementation was corrected; test_manager_stream.py went red on the upstream count before the fixture and the reconcile loop were fixed. Outcome: pass.

T2 — integration (scenario)

$ uv run --project cli python -m pytest -q cli/tests/test_manager_integration.py \
    cli/tests/test_api_instances_integration.py cli/tests/test_manager_stream_integration.py
15 passed in 9.37s

Ten Gherkin scenarios over two worker apps and a manager app joined in-process; four on a worker; one socket-level scenario with three real python -m the_loop.api.serve processes on ephemeral loopback ports (the manager's stream fans two workers' streams in and resumes per member). Outcome: pass.

T3 — contract

$ uv run --project cli python -m pytest -q cli/tests/test_api_contract_parity.py \
    cli/tests/test_mcp_integration.py cli/tests/test_sdk_docs_parity.py
15 passed in 2.70s

The parity test is parametrised over a worker app and a manager app; one MCP tool list on both roles with list_instances and without the registry writers. Outcome: pass.

T5 / T9 — UI, visual and accessibility

$ cd ui && bun run lint && bun run typecheck && bun run test && bun run build
$ oxlint --type-aware
$ tsc --noEmit
 Test Files  21 passed (21)
      Tests  324 passed (324)
✓ built in 1.27s

Screenshots of the built bundle (bun run preview) pointed at the live manager of the manual walk-through, light and dark, under ui/: instances-*.png (the fleet table, the not-live banner, the Register card), instance-laptop-a-*.png (the pane: identity, daemons with verbs, the config editor mounted with instance=laptop-a), instance-cloud-1-*.png (every control disabled with the reason), board-*.png (chips on rows, the filter, the session panel naming the instance). State words beside every dot, the filter a native <select>, actions as buttons: checked by eye against the locked prototype. Outcome: pass.

T7 — performance, bounded

$ uv run --project cli python -m pytest -q cli/tests/test_manager_fleet.py cli/tests/test_manager_stream.py \
    -k "hung_member or upstream_per_live_member"
2 passed, 28 deselected in 2.46s

A hung member costs at most the timeout (elapsed < 4 s with timeoutSeconds: 1); four subscribers hold exactly one upstream per live member. Outcome: pass.

T8 — security / abuse cases

$ uv run --project cli python -m pytest -q cli/tests -k "foreign_url or mismatched_member or malformed_member \
    or hung_member or ambiguous_ref or unknown_instance or foreign_instance or proxied_config \
    or overwrites_a_members_claim or nothing_of_the_caller or refuses_to_boot \
    or stamp_origin or hand_edited_entry"
16 passed, 4968 deselected in 6.47s

One negative test per abuse case A1–A9; the mapping is in security-review.md. The two selectors added last cover self-review round 3's C9 (the subject of a fleet event is kept, never confused with its origin) and C14 (a hand-edited registry entry is refused by index before a write). Re-run after round 3 (0b2cd86). Outcome: pass.

T10 — migration / upgrade

$ uv run --project cli python -m pytest -q cli/tests/test_config_schema_parity.py cli/tests/test_docs_parity.py \
    cli/tests/test_migrations.py cli/tests/test_instance_integration.py cli/tests/test_api_routers_integration.py \
    cli/tests/test_eventlog.py
104 passed in 3.50s

A config without role is a worker and every pre-existing suite passes unchanged (T12 below); the packaged schema is byte-identical to the authored one; CURRENT_CONFIG_VERSION is unchanged (0.10.0); a member without role in its document probes as live (test_a_member_with_no_role_field_probes_as_a_worker); the docs↔schema and SDK↔docs parity tests pass. Outcome: pass.

T11 — manual exploratory

The procedure and its output: manual.md. Outcome: pass.

T12 — the repository's own gates

$ uv run ruff check cli hooks
All checks passed!
$ uv run ruff format --check cli hooks
388 files already formatted
$ uv run pyright cli
0 errors, 0 warnings, 0 informations
$ npx markdownlint-cli2@0.18.1 "**/*.md"
Summary: 0 error(s)
$ python scripts/validate_config.py
VALID   .the-loop/collaborators.yaml
VALID   skills/the-loop/templates/collaborators.yaml
VALID   .the-loop/cli-config.yaml
VALID   skills/the-loop/templates/cli-config.yaml
$ cd cli && uv run python -m pytest -q          # the CI form of `make test`
4965 passed, 1 skipped, 2 warnings in 189.92s (0:03:09)     # before the review rounds
4983 passed, 1 skipped, 1 warning in 196.41s (0:03:16)      # after self-review round 3
$ uv run --project cli python -m pytest -q cli/tests   # from the repository root
4965 passed, 1 skipped, 1 warning in 213.81s (0:03:33)

make validate and make test are run through their underlying commands: the container's uv on PATH is 0.8.17 and the Makefile pins >=0.12; a 0.12.21 uv from PyPI ran every command above. Outcome: pass.

T13 — the security review gate

security-review.md: the abuse-case table with verdicts, the harness's security-review skill run over the diff, and the tier-4 sign-off line for the owner.