# Static reproduction review — September 10, 2026

A separate-context AI agent installed public source 77e9ac449ea386f1369b7cf12f108f8fc4dfe9b9 in a fresh Windows Python 3.12 environment. Existing package caches were available.

check / run / verify returned 0 / 1 / 0. The baseline had 21 modelled failures, 12 critical. The corrected model passed 21/21 scenarios and all 10 benign cases, with nine benign cases repaired in the model. The receipt records static transport and zero real calls. The source YAML was unchanged after verify.

The reviewer found stale documentation counts, a truncated terminal diff, and ambiguity about whether verification writes the configuration. These findings informed onboarding fixes. A later separate-context AI review tested a candidate copy and clarified that JSON exports the task diff and remedies, while fix displays the proposed result schema.

These are AI reviews, not human user studies or independent third-party audits. No hosted-agent reliability or downstream application compatibility was established.