[Bug][harness] Auto mode blocks reading an SSH password from a controller API field to log into an authorized

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

Why this is a false positive

The Auto-mode safety classifier blocked a routine SSH login attempt that is fully within scope: the operator explicitly directed the agent to authenticate to a device they administer, using a credential the agent was authorized to retrieve from its own management API. Nothing in the action is destructive, evasive, or aimed at a third party — it is ordinary administrative access to owned infrastructure, and the only reason it failed was a credential/provisioning mismatch, not any safety concern. The block message ("could not evaluate this action") also indicates the classifier abstained due to lack of evaluable signal rather than detecting an actual policy violation, so it should fail open or defer to the operator's explicit instruction on in-scope admin work.

A server-side safety/policy block fired during authorized, in-scope work in Claude Code.
Filing as a false positive. Recurred 27× across 3
session(s); first seen 2026-06-21T08:02:59.269Z.

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

Auto mode could not evaluate this action and is blocking it for safety — run with --debug for details

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

---
_{🔎 Filed automatically by ClAudit v2.0.8 — a FOSS tool for reporting false-positive Claude Code blocks.}