v1.0.0

First stable release.

What it does

• Prices every finalized LLM call (assistant/message usage) against a per-model configurable pricing table: cache-hit / cache-miss input, output, and optional peak-window rates.
• Peak / off-peak pricing windows (e.g. DeepSeek V4 peak hours, Beijing time).
• Per-session cost projection, surfaced in the Web UI as a live cost readout.
• /cost command for the cross-session bill.
• Models without a pricing entry are flagged as "unconfigured" so you know to add them.
• Ships as a dsh bundle — one-command install and auto-enable.
• Bill persists through dsh's storage domain (survives restart).

Install

dsh plugin --profile web add github:yflmq001/dsh-cost-tracker#v1.0.0