[Suggestion / Feature] Make Max / Coding Plan limits transparent and overload-safe instead of silent restrictions

[Suggestion / Feature] Make Max / Coding Plan limits transparent and overload-safe instead of silent restrictions

Verified context (from Z.AI's own Fair Usage messaging):
When usage exceeds the Fair Usage Policy on the Coding Plan, users get:

"Your account's current usage pattern does not comply with the Fair Usage Policy, and your request frequency has been limited... To restore access, please go to the top of the Console → Coding Plan → Personal Package Overview page and submit a request to lift the restriction."

Two problems with the current behavior:

1. No stated end time. The restriction message gives no duration or ETA, so users cannot tell whether they are throttled for minutes or days.
2. Manual lift required. Recovery depends on the user finding a console page and submitting a request — there is no automatic reset on a known schedule.

Separate overload symptom (GLM-5.2 peaks):
Requests return [1305] model currently overloaded (HTTP 529) and the session drops into a "reconnecting…/continue" loop. Reproduced with a 7-minute gap between requests, so it is server saturation, not user request frequency.

Proposal:

1. Transparent caps: Show the user's Fair Usage allowance and current usage in the Console (requests/day or tokens/day), so the boundary is visible before a restriction hits.
2. Stated duration: When a restriction applies, show an ETA or auto-reset time instead of "submit a request to lift."
3. Overload handling: During model saturation, show "model busy, ETA X min" + auto-queue instead of a silent drop to "reconnecting."
4. Soft daily ceiling option: A clear daily budget (e.g. ~50–200 prompts/day for website use) caps worst-case load while staying generous for normal use.

Why this matters: Providers who show explicit, visible limits (e.g. Anthropic) get fewer "why am I blocked" complaints because users understand the boundary. A known limit beats an unknown wall.

Regards,
Roman

• http://www.rommark.dev
• Vibe Coders Telegram Channel: https://t.me/VibeCodePrompterSystem

No repository files, tests, or entry points are identified. Start by locating the Console Fair Usage and limits flow, plus the GLM-5.2 overload and reconnect handling, and confirm whether this repository controls them. Done requires an agreed design for visible limits, reset timing, and overload recovery, with implementation locations and acceptance tests identified.