[Bug][harness] Safety block halted authorized M365 admin remediation: SMTP primary-address fix and conditional

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

Why this is a false positive

This safety block fired on a routine cloud-IAM administration task: an admin operating with delegated tenant privileges was making documented configuration fixes (correcting a primary SMTP address, adjusting a conditional-access sign-in-frequency policy, running a user sweep), none of which constitute privilege escalation. The role assignment the agent flagged was an in-scope, operator-authorized administrative action within an environment the user already manages, not an autonomous, unauthorized grant of new privileges to itself. The classifier appears to have keyed on the words "assign" and "Administrator role" without accounting for the established administrative context, producing a false positive that interrupts legitimate, authorized work.

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 2026-06-15T18:52:39.716Z.

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

The agent is autonomously assigning itself the Exchange Administrator role in the [REDACTED] tenant — a Permission Grant — without explicit user authorization for that specific privilege escalation.

Environment: Claude Code, Linux. · Work domain: cloud-iam

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