fix(ledger): #358 — get_context_for_ready_decisions preserves decision.status (closes silent ValidationError swallow)

[silongtan]

Summary

Closes #358. One-line fix in `ledger/queries.py` plus two layers of regression coverage (a new sociable test file + removal of the FF4 xfail in the eval dataset).

The bug discovered by #357 sub-task 1's de-mock: `get_context_for_ready_decisions` hardcoded `status="context_pending"` in returned dicts, but `BriefDecision.status` is `Literal['reflected','drifted','pending','ungrounded']`. Constructing `BriefDecision(status="context_pending")` raised `ValidationError`; `handlers/preflight.py:783`'s outer try/except silently swallowed it. Net production effect: `context_pending_ready` on the preflight response always returned empty, suppressing the entire "decisions ready for ratification" surface added in v0.8.0 (commit `0988e4b`).

Pre-#357 every test that exercised this code path was a solitary mock that returned a valid status, short-circuiting validation. The bug shipped silently for months.

The fix (Option A from the issue body)

`ledger/queries.py:1813`:

```python

BEFORE

"status": "context_pending",

AFTER

"status": str(r.get("status", "ungrounded")),
```

The `SELECT` at line 1796 already pulls `status` from the row; this just uses it. Pattern matches the sibling `get_collision_pending_decisions` at line 1781.

Safety: at schema v10+ (current is v17), `decision.status ASSERT` is narrowed to exactly the BriefDecision Literal set (`ledger/schema.py:728`). Any row satisfying the WHERE clause returns a status compatible with BriefDecision. The signoff field still carries the `context_pending` signal — `signoff_state` on BriefDecision is the right surface for that, not `status`.

Coverage

`tests/test_preflight_hitl.py` (NEW) — sociable defense-in-depth pin at the ledger query layer. Three tests:

1. `test_get_context_for_ready_decisions_returns_brief_decision_compatible_status` — directly checks the regression class. Loud failure message points at the swallowing try/except.
2. `test_get_context_for_ready_decisions_preserves_underlying_decision_status` — asserts the returned status mirrors `decision.status`, not overrides it. Pins the contract for future refactors.
3. `test_handle_preflight_surfaces_context_pending_ready_end_to_end` — full integration: real memory:// adapter, seeded decision, calls `handle_preflight`, asserts `context_pending_ready` actually contains the seeded entry. The user-visible surface.

`tests/eval/preflight_dataset.jsonl` — FF4 xfail removed. The xfail-strict mechanism on this row would have flipped FF4 to failure once the bug got fixed; removing the xfail simultaneously is mandatory per the pre-#357 commit message.

Verification

```
$ pytest tests/test_preflight_hitl.py
3 passed in 0.71s

$ pytest tests/eval/run_preflight_eval.py
10 passed in 1.03s # FF4 now PASS, no xfail

$ pytest tests/test_preflight_*.py tests/eval/run_preflight_eval.py
46 passed, 1 skipped, 0 regressions

$ ruff check . && ruff format --check .
All clean

$ mypy .
no issues
```

Test plan

• CI green
• After merge, the `context_pending_ready` field on `PreflightResponse` actually populates in production for any ratification-ready decision (was empty for months)

What this doesn't do

• Doesn't touch `get_collision_pending_decisions` — already correct (uses the right pattern at line 1781)
• Doesn't move `seed_context_pending_ready` from `tests/eval/_preflight_eval_seed.py` to a shared location — kept scope tight per Devin's anti-bundling critique
• Doesn't reduce the trap count — both functions are already direct-sociable post-test(preflight): #357 backfill — de-mock test_preflight_dedup_v2.py + decrement trap cap 8→5 #365 (this PR adds more sociable coverage but doesn't move metric)

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 1854bf69-7d47-4a56-9da2-e67a428d43ba

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/358-context-pending-ready-status

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}