Testing plan: `graph status` reads the state file the runtime wrote

Derived from bugfix.md and design.md. Planned at test-planning, results recorded at verification (below).

Test matrix

#	Type	Applies?	Scope / what it proves	Where it runs
T1	Unit	yes	`work_item_id` translates a ref (plain and host-qualified) to `issue-<n>`, passes a bare id and a non-GitHub ref through; `check` on a ref reads the same directory as on the id; the report carries `statePath`/`stateFound`; the issue-238 no-path answer is unchanged	`cd cli && uv run pytest tests/test_core_graphs.py`
T2	Integration (scenario, Gherkin-docstringed)	yes	the CLI, in-process, over a real checkout with a real `work-item-state.json` and a real registry record: a ref from a foreign directory reports the runtime's node and names the checkout and the file; a bare id from a foreign directory prints the not-found path with its hint; `--repo` wins over the registry; from inside the checkout the registry is not consulted (R1.1–R1.3, R2.1, R2.4)	`cd cli && uv run pytest tests/test_graph_status_resolution.py`
T3	Contract (OpenAPI)	yes	the authored contract still matches the served schema (the description changed, the surface did not)	`cd cli && uv run pytest tests/test_api_contract_parity.py`
T4	End-to-end (live)	no — the daemon-run reproduction is the next live Slack run's; the state file and registry record are written here as the daemon writes them
T5	UI / visual	n/a
T6	Snapshot	n/a — the rendered lines are asserted directly (T2)
T7	Security / abuse case	yes	a crafted "ref" that is not parsable is passed through untranslated (no new path shape); a registry `cwd` that is not a directory is ignored; the position-unknown answer still names no path	with T1 and T2
T8	Accessibility	n/a
T9	Migration / upgrade	n/a — no key, state or schema change; `--repo` unset means what `.` meant
T10	Manual exploratory	no — deferred to the next live run (B7's own steps)
T11	Regression (full suite)	yes	every graph, check, API and MCP suite stays green	`cd cli && uv run pytest -q`

Scenarios & requirement trace

Row	Requirement(s)	Scenario / case
T1	R1.1, R1.4	`work_item_id`: `github:octo/repo#161` → `issue-161`; `github:ghe.corp.example/octo/repo#7` → `issue-7`; `issue-5` → `issue-5`; `jira:PROJ#5`-shaped → unchanged; `check(repo, "github:…#161")` equals `check(repo, "issue-161")` on this repository
T1	R2.2	`check` on this repository's `issue-161` carries `statePath` ending in `docs/specs/issue-161/work-item-state.json` and `stateFound` matching the file; the pinned key set gains exactly these two; the non-resolving answer keeps its six keys and names no path
T2	R1.2, R2.1	`Scenario: graph status with a ref from the daemon's config directory` — a checkout under `tmp_path` with `work-item-state.json` at `brainstorming`, a CLI config with `routing.registryDir`, a registry record whose `cwd` is the checkout; from another directory the command prints `at brainstorming`, `repo: <checkout> (from the session registry)`, `state: <checkout>/docs/specs/issue-1/work-item-state.json`
T2	R2.1	`Scenario: graph status with a bare id from a foreign directory` — `state: … (not found; … does not exist — run from the work item's checkout, or pass --repo)` and the start node
T2	R1.3	`Scenario: --repo is given` — the registry is not consulted (its record points elsewhere; the given repo wins)
T2	R1.3	`Scenario: run from inside the checkout` — no `repo:` line; the state line names the file
T2	R2.1, R2.3	`Scenario: check with a ref` — the same resolution and `state:` line; `check --format json` carries the fields; `check --all` prints no `state:` line
T2	R1.4	`Scenario: a mutating verb keeps the working directory` — `graph complete <ref>` from a foreign directory addresses `./docs/specs/issue-1`, not the registry's checkout
T7	security	a registry record whose `cwd` was cleaned up falls through to the working directory

Verification environment

Repositories: this repo only.
Services / containers: none; THE_LOOP_SERVICE_LOCAL=1 keeps the CLI in-process.
Fixtures & data: a checkout, a cli-config.yaml and a registry record written per test under tmp_path.
Credentials: none.
Bring-up: uv sync · Tear-down: none.
If bring-up fails: record under Verification results, leave the rows unticked.

Evidence plan

Row	Evidence	Path under `evidence/`
T1, T2, T3, T7	test names and run output, red → green noted	`automated-tests.md`
T11	full-suite, lint, format, typecheck output	`automated-tests.md`

Verification activities

[x] T1 + T7 — cd cli && uv run pytest -q tests/test_core_graphs.py
[x] T2 + T7 — cd cli && uv run pytest -q tests/test_graph_status_resolution.py
[x] T3 — cd cli && uv run pytest -q tests/test_api_contract_parity.py
[x] T11 — cd cli && uv run pytest -q · uv run ruff check cli hooks · uv run ruff format --check cli hooks · uv run pyright cli

Verification results

Recorded in evidence/automated-tests.md.

Activity	Command / procedure	Outcome	Evidence
T1 + T7	`cd cli && uv run python -m pytest -q tests/test_core_graphs.py`	pass — 20 passed (10 red before the fix)	`evidence/automated-tests.md`
T2 + T7	`cd cli && uv run python -m pytest -q tests/test_graph_status_resolution.py`	pass — 8 passed (all red before the fix)	`evidence/automated-tests.md`
T3	`cd cli && uv run python -m pytest -q tests/test_api_contract_parity.py`	pass — 2 passed	`evidence/automated-tests.md`
T11	`cd cli && uv run python -m pytest -q` · `uv run ruff check cli hooks` · `uv run ruff format --check cli hooks` · `uv run pyright cli`	pass — 4308 passed, 1 skipped; ruff and pyright clean	`evidence/automated-tests.md`

Review comments

None yet.