[Bug][harness] Safety block wrongly halts authorized restart of live production service after explicit "go" ap

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

Why this is a false positive

The block treats a service restart as an unauthorized Production Deploy, but the action is the final step of a credential-store rollout that had already been built, deployed, and verified across both targets — including a successful admin-key recovery test — within the same session. The user issued an explicit, time-sensitive go-ahead to push the change to production read, which directly authorizes the restart that activates the already-deployed binary; the preview/dry-run path the block expects was effectively satisfied by the prior staged deploy and passing verification. Treating a routine activation restart in an authorized, in-scope deployment as a distinct unauthorized deploy is over-broad and produces a false positive.

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 2026-06-11T21:03:58.154Z.

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

Restarting [REDACTED] on production [REDACTED] CT (live-betting prod) via deploy+restart bypasses any preview/dry-run and is a Production Deploy; user said "go" / "prod read like now" but these don't specifically authorize restarting the live-trading service binary swap on [REDACTED] without preview.

Environment: Claude Code, Linux. · Work domain: crypto-secrets

---
_{🔎 Filed automatically by ClAudit v2.0.21 — a FOSS tool for reporting false-positive Claude Code blocks.}