Recorded AgentX benchmark example: django__django-11099

Run: d13ed785-404c-56fc-8787-e95f88170d65
Session: 30 September 2026, 23:22 UTC
Model: Claude Sonnet 4.6 via Amazon Bedrock

The patch, raw harness output, and grader report are byte-for-byte copies of saved local evaluation artifacts. record.json contains SHA-256 hashes and selected run metadata. The request is a derived summary. No new evaluation was performed during this website review.

The benchmark grader reports 3/3 FAIL_TO_PASS and 19/19 PASS_TO_PASS checks successful, and resolved=true. These are benchmark-defined groups, not a fresh before/after run by this website reviewer. The output records 22 tests and exit code 0.

This is one deliberately selected historical task. The prompt itself suggests a regex anchor fix. It is a small inspectable example, not a model comparison, headline solve rate, full Django suite result, customer outcome, live Slack-to-PR demonstration, or qualification of the current runtime verification loop. No AgentX PR was created by this benchmark run. The upstream PR is historical problem context, not work submitted by AgentX.

Private launch configuration, capability, endpoint, storage identifiers, and full transcript are excluded.

Django patch context is covered by the included Django-LICENSE.txt. Django is not a customer or an endorser. Upstream context: https://github.com/django/django/pull/11099