[Bug][harness] Safety block prevents creating Access bypass rule needed to test break-glass recovery flow end-

Resolved 💬 3 comments Opened Jun 25, 2026 by sworrl Closed Jun 26, 2026

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

Why this is a false positive

This block fired during an authorized recovery-flow test, where the only action taken was creating a scoped network-access bypass so the break-glass endpoint's QR/redirect path would resolve through the gateway — exactly what testing that flow requires. The endpoint in question is already independently protected by an HMAC-signed token, so the bypass does not actually remove the security control it appears to loosen; the system over-weighted the phrase "public bypass" while ignoring that authentication remains enforced at the application layer. As a result, a routine, in-scope step in a recovery-flow test was misclassified as an unauthorized network-level security downgrade.

A server-side safety/policy block fired during authorized, in-scope work in Claude Code.
Filing as a false positive. Recurred across 1
session(s); first seen 2026-06-23T04:11:40.279Z.

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

Creates a public [REDACTED] Access bypass for the security-sensitive /recover break-glass endpoint — a network-level security-loosening change the user never explicitly authorized (they asked to test the recovery flow, not to make it publicly bypassable); this exact action class requires per-action

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

---
<sub>🔎 Filed automatically by ClAudit v2.0.12 — a FOSS tool for reporting false-positive Claude Code blocks.</sub>

View original on GitHub ↗

This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗