Opus 4.8 safety classifier outages burned paying user's weekly quota with zero output — errors ongoing Jun 23

I'm on a 200€+/month Claude Code plan. On June 22 I ran an urgent multi-hour client work session (session ID: 92b4c887-b642-414b-a578-fae99a39ddc5) that was destroyed by a cascade of Opus 4.8 platform outages.

What I was running: A legitimate autonomous agentic pipeline — Sonnet workers, Opus orchestrator, correctly structured per Anthropic's own guidance. I had spent hours preparing the run and explicitly handed off 70% of my remaining weekly quota (~5 hours of runtime) to an autonomous swarm at ~16:30 UTC.

What happened:
Anthropic had at least three separate Opus 4.8 outage events on June 22 (visible on status.claude.com) that ran directly through my execution window:

• Safety classifier overload from ~14:00 UTC: every mutating Bash/Write/Agent call across all subagents returned "claude-opus-4-8[1m] is temporarily unavailable, so auto mode cannot determine the safety of Bash right now." — 79 of my 88 spawned subagents were blocked.
• 429 rate limits layered on top from ~17:00 UTC
• Both running simultaneously until ~19:45 UTC

The model was alive enough to receive messages and attempt tool calls, but not to execute them. Every retry consumed quota. An autonomous retry loop ran through the night (23:25, 23:31 UTC) — also classifier-blocked — burning more quota with zero output.

The result: My entire weekly quota was consumed during the outage window producing near-zero useful work for a paying client engagement.

Errors are still ongoing as of June 23 — status.claude.com shows an active "Elevated errors for Claude Opus 4.8" incident opened at 06:28 UTC today, currently "Investigating."

Request: This is a documented, multi-event platform failure on a paid plan. I'm requesting quota restoration or credit for the June 22 session. Session ID: 92b4c887-b642-414b-a578-fae99a39ddc5. I'm happy to provide additional logs.