# Archive Admission Inspector Standalone offline ZIP inventory, path/refusal report and bounded CRC inspection. Python 3.10+ standard library, Tkinter for the native UI. No other purchases. Run `python app.py`; load input JSON with archive path and optional expectedSha256, then inspect into a new report. CLI: `python checker.py input.json new-report.json`. Generate sample with `python examples/make_sample.py`, then use examples/input.json. Use absolute archive paths or paths relative to the product folder in the UI. Existing reports are refused. Test: `python -m unittest discover -s tests -v`. Reports contain exact SHA-256, inventory, declared/checked expansion, individual findings and admitted boolean. A successfully generated report can still say admitted=false; always check that field. CRC checks occur only after metadata and optional digest admission. No bytes are extracted and no source file is modified. Bounds: 64MiB ZIP, 4096 entries, 128MiB expanded, 200:1 per-entry ratio. Stored or deflated ZIP only. Unsafe/original paths, case duplicates, file/directory collisions, Windows reserved names, links/special entries, encryption and unsupported methods are refused. CRC/header failure after admission is reported as a refusal. Read bytes are counted independently of advertised totals. The entire bounded ZIP is read once into memory, ensuring the hash and inspected bytes refer to the same snapshot. Consumption must recheck that digest; this report cannot prevent later file replacement. Hash equality is not trusted provenance. No antivirus/content moderation, nested archive inspection or extraction feature. Admission does not mean payloads are harmless. Structural tests pass separately from native visual acceptance, which remains pending.