dtzp555-max/ocp v3.8.0

v3.8.0 — Per-Key Quota + Response Cache

on GitHub

What's New

Per-Key Quota (Budget Control)

• Set daily/weekly/monthly request limits per API key
• HTTP 429 with structured error when exceeded
• PATCH /api/keys/:id/quota — partial update
• GET /api/keys/:id/quota — view limits + usage
• Only successful requests count; admin/anonymous exempt

Response Cache

• Opt-in via CLAUDE_CACHE_TTL env var (default 0 = disabled)
• SHA-256 hash of model + messages + temperature/max_tokens/top_p
• Streaming and non-streaming support
• Multi-turn sessions never cached
• GET /cache/stats + DELETE /cache admin endpoints
• Runtime-tunable via PATCH /settings
• Auto-cleanup every 10 minutes

Bug Fix

• SQLite datetime format mismatch with JS .toISOString() — same-day comparisons were broken

Code Quality

• Route conflict fix for DELETE /api/keys/:id vs /quota
• Quota check: single SQL query instead of 3 separate COUNTs
• findKey() encapsulation replaces raw getDb() in server.mjs
• 24-test integration suite (test-features.mjs)

Full Changelog: v3.7.0...v3.8.0