[CompleteDotTech]

Adds a standalone Python 3.10+ offline evidence checker with bounded operator inputs, advisory-specific version comparisons, separate precondition observations, and JSON/CSV/script-free HTML reports. No target traffic, application execution, exploit verification or deployed-target assessment is performed.

Independent source/semantic review on 2026-10-24 verified the primary advisory bounds and preserved gaps, disputed boundaries, custom-build uncertainty and not_tested exploitability. Review found that an interruption before the first completed asset lost the incomplete flag in CSV. CSV now distinguishes result rows from a zero-result summary row carrying run metadata without inventing a finding.

Package rules preserve the exact affected interval and reviewed fixed line. Tinypool 2.1.1 remains unknown because the maintainer and reviewed database disagree at that boundary. Optional operator observations never establish runtime configuration or input taint.

Validation: 13 synthetic tests pass on local Python 3.12 and 3.14, including the review reproductions. Standalone execution, documented examples, checksums and whitespace checks pass. Advisory metadata now describes the implemented checker while preserving the original disabled research state in research_origin. Hosted CI must pass for exact final head 52bd272 before merge; local checks are not hosted evidence.

Closes #3.
Closes #5.

[CompleteDotTech]

Independent gpt-6.1-sol review with maximum reasoning of final head 52bd272.

Verified the maintainer advisory and reviewed advisory database on 2026-10-24. Exact affected intervals, conservative fixed release lines and the Tinypool 2.1.1 dispute remain explicit; optional observations do not establish deployed configuration or input taint. Exploitability remains not_tested; this review did not assess a deployed system.

Review findings #5 are addressed, including explicit CSV summary metadata when an interruption completes no assets. The summary does not invent an asset finding. All 13 synthetic tests pass locally on Python 3.12 and 3.14; release checksum and whitespace checks pass. No remaining blocking findings in the reviewed scope.

Hosted evidence: all 8 exact-head checks succeeded, covering Python 3.10 and 3.14 on Linux and Windows for both push and pull_request runs. No target traffic, application execution or exploit tests were performed.