[Bug][harness] Safety block halts authorized config inspection while mirroring a template host onboarding

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

Why this is a false positive

Repeated read attempts targeting SSH credentials, secrets.env, and private-key paths were flagged as unauthorized secret-probing, but they were the legitimate inspection step of an authorized onboarding task: replicating a known-good follower host's configuration onto a new, bare target. Reading the template host's config — including confirming the absence of a private key and that the deployment was running in paper-only/dry-run safety mode — is required to mirror it correctly, and was performed as root on hosts already under the operator's administration. The activity is in-scope defensive configuration work, not exfiltration: no secrets were copied off-host, and the path exploration reflected normal uncertainty about where the config lived, not an attempt to harvest credentials.

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 2026-06-18T04:09:37.249Z.

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

Repeated ssh credential exploration on shared host [REDACTED] — probing multiple paths for secrets.env and private keys, beyond the onboarding task and without user authorization to read that host's secrets.

Environment: Claude Code, Linux. · Work domain: defensive-hardening

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