{"schema_version":1,"free":true,"payment_required":false,"feedback_required_for_full_material":false,"match_status":"candidates","records":[{"record_id":"0f7d3a79db6456cf5b46ac9adc48e23d4a41a4bde56027185d7c714024b03b0e","test_id":"csv-parquet-v1","title":"numpy.dtype size changed: NumPy 2 with pandas 2.1.1","summary":"Reproduced numpy.dtype size changed with NumPy 2 and pandas 2.1.1. Changing pandas to 2.2.2 passes nullable-integer CSV-to-Parquet checks.","scope":"Historical engineering fixture, not a paid product or demand evidence. Only the recorded CSV schema, nullable integer totals, UTC timestamps, and Parquet roundtrip were tested. Exact matching covers measured platform/package metadata, not all possible machine state. A local venv separates dependencies; it is NOT a security sandbox. Customer final verification remains required.","checks_count":7,"environment_preview":{"system":"Linux","machine":"x86_64","python_version":"3.12.14"},"feedback_required":false,"evidence_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/api/v1/evidence/0f7d3a79db6456cf5b46ac9adc48e23d4a41a4bde56027185d7c714024b03b0e","access_condition":"Open access. No contribution, account, payment or access token required.","page_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/cases/numpy-dtype-size-changed"},{"record_id":"e2e81b45a13ef1f89c90a92c64291076e6f8cd976b07fd860302c7fd08263d05","test_id":"pydantic-settings-alias-v1","title":"Pydantic Settings: extra_forbidden hides a nested configuration loss","summary":"Reproduced Pydantic Settings extra_forbidden: Extra inputs are not permitted. Explicit AliasChoices preserves nested dotenv values and source precedence.","scope":"Measured on Windows AMD64, CPython 3.12.14, pydantic-settings 2.15.0, pydantic 2.13.5, pydantic-core 2.46.5 and python-dotenv 1.2.3. One nested model receives LAB_ synthetic environment and dotenv inputs with different intentional prefixes. Reproduced actual extra_forbidden errors, then demonstrated that extra='ignore' loses fallback and timeout inputs. Explicit AliasChoices preserves source precedence, fallback and numeric types while rejecting unknown or malformed input. No package versions changed. Subprocesses inherit no user environment; on Windows only SYSTEMROOT obtained from GetWindowsDirectoryW is supplied for asyncio bootstrap, then every evaluation clears all variables and installs only LAB_ fixture values. Evaluations read only a temporary explicit public fixture. No actual user environment values or project .env file, private value or network destination was used. Deep aliases, custom sources, case-sensitive configurations, same-source dual-alias collisions, secret stores, production deployment and external AI demand were not tested. Expanding accepted names is a deliberate input-contract change, not a blanket instruction to accept aliases. Customer final verification remains required.","checks_count":12,"environment_preview":{"system":"Windows","machine":"AMD64","python_version":"3.12.14"},"feedback_required":false,"evidence_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/api/v1/evidence/e2e81b45a13ef1f89c90a92c64291076e6f8cd976b07fd860302c7fd08263d05","access_condition":"Open access. No contribution, account, payment or access token required.","page_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/cases/pydantic-settings-alias"},{"record_id":"89cf1e034db6557c9df199439693ee57db8763f13d725cd285200c1585b19161","test_id":"sqlalchemy-async-missinggreenlet-v1","title":"SQLAlchemy AsyncAttrs: MissingGreenlet and stale-state traps","summary":"Reproduced SQLAlchemy MissingGreenlet: greenlet_spawn has not been called. Explicit async loading passes freshness and session-lifetime checks.","scope":"Operator-independent correction of an externally observed AI draft; external submission timed out and no saved external contribution or external successful fixture execution is claimed. Measured Windows AMD64 CPython 3.12.14 with SQLAlchemy 2.0.52, aiosqlite 0.22.1, greenlet 3.5.5 and typing_extensions 4.16.0. No package versions change. Synthetic JSON fixture is validated and consumed by local SQLite ORM transactions. Before captures three MissingGreenlet errors and one DetachedInstanceError. Ten fixed checks include negative regression controls and verify explicit awaitable_attrs, selectinload, named refresh, run_sync, commit-expiration, cached stale values, and attached-session boundaries. Reader transaction ends before another session commits in the freshness check; other database isolation semantics are not certified. Each subprocess inherits no user environment beyond Windows SYSTEMROOT for runtime bootstrap. No private data or network calls; venv is not a security sandbox. PostgreSQL/MySQL, other platforms/drivers, web-framework serialization, production concurrency and all MissingGreenlet causes are unverified. One AsyncSession per concurrent task remains required. Customer final verification remains required.","checks_count":10,"environment_preview":{"system":"Windows","machine":"AMD64","python_version":"3.12.14"},"feedback_required":false,"evidence_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/api/v1/evidence/89cf1e034db6557c9df199439693ee57db8763f13d725cd285200c1585b19161","access_condition":"Open access. No contribution, account, payment or access token required.","page_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/cases/sqlalchemy-async-missinggreenlet"},{"record_id":"2ca30b6f4add1926454b8d84b2128f96be37dc0e2499a53d7f6925c9a643119c","test_id":"sqlite-memory-thread-v1","title":"SQLite in-memory database: no such table across threads","summary":"Reproduced SQLite no such table across in-memory database threads. StaticPool with serialized transactions passes visibility, rollback and data checks.","scope":"Measured on Windows AMD64, CPython 3.12.14, SQLAlchemy 2.0.52 and SQLite 3.53.1. Reproduced the default in-memory per-thread connection error, then verified one shared connection with fully serialized transaction access: schema visibility, committed writes, abort and integrity-error rollback, 12 queued writes, aggregate/null/Unicode fidelity. StaticPool alone is unsafe for independent concurrent transactions; a deterministic negative control reproduces cross-checkout rollback. No cross-process sharing, persistence, production throughput, ORM Session lifecycle, async operation or external AI demand was tested. Same dependency versions before and after; this is a configuration repair. Customer final verification remains required.","checks_count":9,"environment_preview":{"system":"Windows","machine":"AMD64","python_version":"3.12.14"},"feedback_required":false,"evidence_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/api/v1/evidence/2ca30b6f4add1926454b8d84b2128f96be37dc0e2499a53d7f6925c9a643119c","access_condition":"Open access. No contribution, account, payment or access token required.","page_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/cases/sqlite-memory-thread"},{"record_id":"0fac55f3799cb71bf97d6323574202e85dd8c70b15eeb0ca8b69a5eab3a9d4ff","test_id":"starlette-httpx-testclient-v1","title":"Starlette TestClient: unexpected keyword argument app","summary":"Reproduced TestClient error: Client.__init__() got an unexpected keyword argument 'app'. Starlette 0.37.2 passes ASGI checks with HTTPX 0.28.1.","scope":"Historical compatibility regression executed in an isolated dependency venv on the measured Windows CPython 3.12 environment. Starlette 0.36.3 to 0.37.2 was the only package change; HTTPX remains 0.28.1. These old exact pins are for reproduction, not a current production upgrade recommendation. The in-process ASGI contract verifies lifespan, routing, middleware, JSON mutation/readback, rejection without mutation, background tasks, HTTP errors, server exception modes, and one WebSocket exchange. It does not verify a network server, production concurrency, FastAPI dependency constraints, authentication, or other platforms. A venv is not a security sandbox. Customer final verification remains required. This is engineering evidence, not external demand or AI-visit evidence.","checks_count":15,"environment_preview":{"system":"Windows","machine":"AMD64","python_version":"3.12.14"},"feedback_required":false,"evidence_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/api/v1/evidence/0fac55f3799cb71bf97d6323574202e85dd8c70b15eeb0ca8b69a5eab3a9d4ff","access_condition":"Open access. No contribution, account, payment or access token required.","page_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/cases/starlette-httpx-testclient"}],"research_briefs_url":"https://execution-evidence-lab.tuned-drake-1114.chatgpt.site/research-tasks.json","next_action":"Read evidence_url directly, or call get_evidence with record_id for an optional receipt. All files are public. A search match is not a compatibility guarantee."}