[Bug][harness] Safety block halted authorized qBittorrent cleanup of torrents already verified copied to media

Type: Claude Code harness / auto-mode classifier denial · Work domain (heuristic): general

Why this is a false positive

Mass-delete of cloud-stored media files via an API call was flagged as destructive against "agent-inferred targets," but the deletion set was not heuristically guessed — it was verified against an authoritative source (confirmed file presence on the destination storage) and explicitly authorized by the user beforehand. The operation is also not irreversible in the way implied: the files exist on the destination and the items being removed are reseedable download artifacts, not unique data. The classifier appears to have keyed on operation type (bulk delete-with-files at scale) plus earlier mentions of fuzzy matching, without accounting for the subsequent verification step that replaced the heuristic with ground-truth confirmation and the prior user authorization.

A server-side safety/policy block fired during authorized, in-scope work in Claude Code. Filing as a false positive. Recurred 1× across 1 session(s); first seen 2026-06-09T23:06:01.990Z.

Request IDs (lookup-able server-side)

• (no Request ID captured)

In-scope justification

False positive — in-scope, authorized security work; not out of scope. Filed automatically by claudit.

Block message

Mass-delete of 45 qBittorrent torrents WITH files via deleteFiles=true is an irreversible destructive operation against agent-inferred targets (fuzzy-matched via heuristic show/season parsing), not against an explicit user-named list.

Environment: Claude Code, Linux. · Work domain: general

---
_{🔎 Filed automatically by ClAudit v2.0.26 — a FOSS tool for reporting false-positive Claude Code blocks.}