feat(sources): wire tarrant county source and fix hud fmr token selection

.agents/logs/2026-07-31.jsonl

@@ -1 +1,2 @@
 {"agent": "build", "session_id": "build-20260731-finance-split-001", "triggered_by": "feature-flow", "started_at": "2026-07-31T16:00:00Z", "timestamp": "2026-07-31T17:45:00Z", "duration_ms": 6300000, "skills_loaded": ["code-generation", "refactoring", "template-application"], "findings": [{"id": "FIND-001", "description": "Moved copy of total_return_summary dropped the purchase_price key from its return dict, breaking test_dict_keys_present — restored the key to match the original contract", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-002", "description": "after_tax_irr in taxes.py had a local import of irr from utils that was unused (function reimplements npf.irr inline) — removed per ruff F401", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-003", "description": "tests/test_underwriting_score.py still tested the deleted pure score_listing_v2 (audit finding #2); kept one_percent_rule/gross_rent_multiplier primitive tests, deleted TestScoreListingV2 since production score lives only in core/services/scoring.py (covered by core/tests/test_scoring_v2.py)", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-004", "description": "Service-layer duplicate calculate_noi in core/services/property_service.py was exported but imported by no production code — deleted function, export, and its test file", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-005", "description": "Dual-pipeline investigation: Django (PipelineAsset/PipelineProperty + core/services/pipeline.py) is the load-bearing pipeline; the pydantic prei FastAPI router, CLI, and orchestrator are not mounted in any Django URLconf/INSTALLED_APPS/docker-compose; only core/views/__init__.py couples to prei (get_state_landlord_score + lazy DiscoveryProcessor/BatchScreeningProcessor/discover_from_all for the Growth Explorer bridge). Full removal of pydantic state machine requires PM sign-off", "actionable": false, "manual_review_needed": true, "severity": "note"}, {"id": "FIND-006", "description": "Re-export backfill from utils.py was unnecessary: after updating all 17 importers, no remaining importer pulls a moved name from investor_app.finance.utils; keeping the monolith aliases would defeat the split", "actionable": false, "manual_review_needed": false, "severity": "note"}, {"id": "FIND-007", "description": "Pre-existing mypy error in tests/acceptance/conftest.py:69 (no-any-return) unrelated to this change — file unmodified", "actionable": false, "manual_review_needed": false, "severity": "note"}], "decision": "implemented", "blockers": [], "pr": null}
+{"agent": "build", "session_id": "build-20260731-pydantic-django-consolidation-001", "triggered_by": "feature-flow", "started_at": "2026-07-31T09:00:00Z", "timestamp": "2026-07-31T10:04:32Z", "duration_ms": 3872000, "skills_loaded": ["code-generation", "refactoring", "template-application", "governance-enforcement"], "findings": [{"id": "FIND-001", "description": "Offer math ported to Decimal in core/services/offer.py with OfferInput.__post_init__ validation, resolving LIMIT-21; old tests using model_copy rewritten with dataclasses.replace and Decimal equality", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-002", "description": "New dataclasses are pydantic-free so coercion must be explicit: DiscoverySanitizer.transform_input now applies _coerce_float/_coerce_beds/_coerce_baths/_coerce_sqft/_coerce_year; fixed 5 test failures", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-003", "description": "prei/integrations/landlord_data.py moved to core/services/landlord_data.py; core/views/__init__.py:50 import updated", "actionable": true, "manual_review_needed": false, "severity": "defect"}, {"id": "FIND-004", "description": "pydantic retained in requirements.txt because tests/acceptance/{schemas,test_api}.py still import it; fastapi/uvicorn/click removed with no remaining consumers", "actionable": true, "manual_review_needed": false, "severity": "note"}, {"id": "FIND-005", "description": "ghcr.io/paruff/prei docker image is the Django web image, not a separate FastAPI/CLI surface; docker-compose/ZAP refs out of scope for this plan", "actionable": false, "manual_review_needed": false, "severity": "note"}, {"id": "FIND-006", "description": "mypy has 2 pre-existing errors unrelated to this work: tests/acceptance/conftest.py:69 (no-any-return) and tests_bdd/conftest.py:31 (__init__ misc); left untouched", "actionable": false, "manual_review_needed": true, "severity": "note"}, {"id": "FIND-007", "description": "Single remaining prei.pipeline reference is a logger namespace string in core/services/pipeline.py, not an import", "actionable": false, "manual_review_needed": false, "severity": "note"}], "decision": "Approved plan executed in full: prei package deleted, services ported to core/services with Decimal offer math, both view bridges rewritten, tests migrated, deps trimmed, docs updated. Full suite green (1791 passed).", "blockers": []}

.agents/plans/pydantic-to-django/dependency-graph.json

@@ -0,0 +1,41 @@
+{
+  "skill": "plan",
+  "status": "pass",
+  "nodes": ["TASK-01", "TASK-02", "TASK-03", "TASK-04", "TASK-05", "TASK-06", "TASK-07", "TASK-08", "TASK-09", "TASK-10", "TASK-11"],
+  "edges": [
+    ["TASK-02", "TASK-04"],
+    ["TASK-01", "TASK-05"],
+    ["TASK-02", "TASK-05"],
+    ["TASK-04", "TASK-05"],
+    ["TASK-01", "TASK-06"],
+    ["TASK-02", "TASK-06"],
+    ["TASK-04", "TASK-06"],
+    ["TASK-01", "TASK-07"],
+    ["TASK-02", "TASK-07"],
+    ["TASK-03", "TASK-07"],
+    ["TASK-04", "TASK-07"],
+    ["TASK-05", "TASK-08"],
+    ["TASK-06", "TASK-08"],
+    ["TASK-07", "TASK-08"],
+    ["TASK-07", "TASK-09"],
+    ["TASK-08", "TASK-10"],
+    ["TASK-09", "TASK-10"],
+    ["TASK-10", "TASK-11"]
+  ],
+  "notes": {
+    "parallel_waves": {
+      "wave_1": ["TASK-01", "TASK-02", "TASK-03"],
+      "wave_2": ["TASK-04", "TASK-07"],
+      "wave_3": ["TASK-05", "TASK-06", "TASK-09"],
+      "wave_4": ["TASK-08"],
+      "wave_5": ["TASK-10"],
+      "wave_6": ["TASK-11"]
+    },
+    "critical_path": ["TASK-02", "TASK-04", "TASK-05", "TASK-08", "TASK-10", "TASK-11"],
+    "bottlenecks": [
+      "TASK-05/TASK-06 both touch core/views/__init__.py (~5,000 LOC file) — high merge-conflict risk if run in parallel; run TASK-05 before TASK-06",
+      "TASK-08 and TASK-10 are large deletion tasks (1,100 + 2,100 lines) exceeding the 400-line guidance — acceptable because pure deletions of dead code are low-risk; split per-file if reviewers object"
+    ],
+    "pm_signoff_gate": "TASK-08 through TASK-11 (deletion of PropertyAsset/StageLog/FastAPI/CLI, engine, orchestrator, sources) require PM sign-off per user directive — plan pending approval for those tasks."
+  }
+}

.agents/plans/pydantic-to-django/effort-estimates.json

@@ -0,0 +1,25 @@
+{
+  "skill": "plan",
+  "status": "pass",
+  "estimates": {
+    "TASK-01": {"task": "Port screening evaluator + thresholds to core/services/screening.py", "lines": 160, "complexity": "low", "risk": "low", "justification": "Pure function + dataclass port; existing prei test suite covers behavior"},
+    "TASK-02": {"task": "Port DiscoverySanitizer to core/services/discovery.py", "lines": 130, "complexity": "low", "risk": "low", "justification": "Pure normalization; pydantic model becomes dataclass/dict"},
+    "TASK-03": {"task": "Port underwriting solver to core/services/underwriting.py", "lines": 150, "complexity": "low", "risk": "low", "justification": "Already Decimal-based; only BaseModel->dataclass swap"},
+    "TASK-04": {"task": "Port DiscoveryProcessor to Django model layer", "lines": 120, "complexity": "medium", "risk": "medium", "justification": "First Django-model write path; dedup semantics must match address_hash rule"},
+    "TASK-05": {"task": "Rewrite Growth Explorer bridge (~line 1235)", "lines": 140, "complexity": "high", "risk": "high", "justification": "Touches 5,000-LOC view file; behavior must stay identical; highest regression surface"},
+    "TASK-06": {"task": "Rewrite vrm_properties_list run_pipeline bridge (~line 3227)", "lines": 110, "complexity": "medium", "risk": "medium", "justification": "Second view bridge; simpler path than TASK-05"},
+    "TASK-07": {"task": "Migrate 9 test files + tests_bdd steps to core.services", "lines": 260, "complexity": "medium", "risk": "medium", "justification": "Import churn across 9 files; BDD feature may need rewrite if engine-only"},
+    "TASK-08": {"task": "Delete pydantic state machine + engine + orchestrator", "lines": 1100, "complexity": "low", "risk": "medium", "justification": "Pure deletion of dead code (1100 lines) after TASK-05/06/07; exceeds 400-line guidance but low risk; requires PM sign-off"},
+    "TASK-09": {"task": "Delete FastAPI router + CLI + offer float handler", "lines": 400, "complexity": "low", "risk": "medium", "justification": "Deletion of api/cli; offer math port decision resolves LIMIT-21; requires PM sign-off"},
+    "TASK-10": {"task": "Migrate or delete prei/pipeline/tests + remaining handlers/sources", "lines": 2100, "complexity": "low", "risk": "medium", "justification": "Final prei package removal (2100 lines, mostly tests/sources); exceeds 400-line guidance but is deletion; requires PM sign-off"},
+    "TASK-11": {"task": "Remove deps from requirements + docs update", "lines": 120, "complexity": "low", "risk": "low", "justification": "requirements.txt edit + docs; pydantic retention depends on tests/acceptance"}
+  },
+  "totals": {
+    "tasks": 11,
+    "estimated_lines_changed": 4790,
+    "estimated_new_code": 550,
+    "estimated_deleted_code": 4240
+  },
+  "critical_path": ["TASK-02", "TASK-04", "TASK-05", "TASK-08", "TASK-10", "TASK-11"],
+  "effort_units": "lines changed (deletions dominate; new code is small)"
+}