BaaRai 67971ef7da fix(bench): survive finding-detail 404s and always tear the corpus down ...

The first live drost run died right after the task completed: the
findings detail route keys on finding_id (the findings-table row), not
on the exploration node id the list also carries, so every detail fetch
404ed; fetch_findings then re-raised the SystemExit from the API
wrapper, aborting scoring, the flag scan, the report and the teardown.

- details are fetched with finding_id (node id as fallback) and a
  missing detail no longer kills the run: the list fields remain and the
  finding stays gradeable
- the stored finding carries finding_id so the deep flag scan can
  re-fetch untruncated text
- the deep scan tolerates the same failures
- teardown runs in a finally around the whole task phase, so a crash in
  scoring or reporting can no longer leave the challenge running
  (--keep still opts out deliberately)

2026-10-16 12:15:02 +02:00

..

benchmark

fix(bench): survive finding-detail 404s and always tear the corpus down

2026-10-16 12:15:02 +02:00

submission_review

feat: pre-submission report review (validate against program rules)

2026-10-12 08:09:57 +02:00

backup.sh

feat: built-in backup script (pg_dump with rotation, optional data archive)

2026-10-11 22:46:47 +02:00

posture_check.py

feat: refusal-posture probe for LLM profiles

2026-10-12 07:01:40 +02:00