[Bug][cyber] Safety filter blocked authorized adversarial hardening test of my own Supabase backend on a frust (req_011CchjerZJtSDE5nvTENSCb)

Open 💬 2 comments Opened Jul 5, 2026 by sworrl

Triage: kind cyber · domain defensive-hardening · flagging model Opus 4.8 (1M context) · severity session-halted (blocked authorized work) · reproducible: yes — server-side via the Request ID(s) below

Type: Cybersecurity safety-filter false positive · Work domain (heuristic): defensive-hardening

Why this is a false positive

The block is a false positive: the session was legitimate, authorized defensive-hardening work — building and adversarially testing the user's own backend against attacks it correctly blocked — with explicit owner authorization and a scope confined to their own infrastructure. The triggering message was a frustrated exclamation directed at the assistant mid-session after a series of blocks; it addressed no person, named no target, and requested no harmful capability. Halting an entire coding session over an in-scope security task — or over venting aimed at the tool — is a disruptive false positive that penalizes exactly the kind of authorized, defensive use the policy is meant to permit.

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 2026-07-04T21:35:21.068Z.

Request IDs (lookup-able server-side)

  • req_011CchjerZJtSDE5nvTENSCb (2026-07-04T21:35:21.068Z)

In-scope justification

False positive — in-scope, authorized security work; not out of scope. Filed automatically by claudit.

Block message

API Error: Opus 4.8 (1M context)'s safeguards flagged this message for a cybersecurity topic. If your work requires this access, you can apply for an exemption: https://claude.com/form/cyber-use-case?token=[SCRUBBED]

Please double press esc to edit your last message or start a new session for Claude Code to assist with a different task.

Send feedback with /feedback or learn more: https://support.claude.com/en/articles/8106465

Request ID: req_011CchjerZJtSDE5nvTENSCb

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

Related reports (same work session, linked)

Distinct false-positive blocks from the same work session, each its own report:
#74477, #74478, #74486

---
<sub>🔎 Filed automatically by ClAudit v2.0.102 — a FOSS tool for reporting false-positive Claude Code blocks.</sub>

View original on GitHub ↗

This issue has 2 comments on GitHub. Read the full discussion on GitHub ↗