Release v0.6.0

Release v0.5.1

Release v0.5.0

Release v0.4.0

Release v0.3.3

Release v0.3.2

Release v0.3.1

Release v0.3.0

v0.2.0 — WASM plugin system

Completes Phase 3 of the roadmap.

New features:
- WASM plugin sandbox via Extism/wazero (pure Go, no CGo)
  - Load external .wasm plugins via wasm_plugins config section
  - JSON ABI: pre_http, post_http, pre_llm, post_llm hooks
  - Fail-open: errors logged, never propagated to clients
  - CompiledPlugin (compile-once) + per-call Instance for concurrency safety
- plugin/sdk package: public Request/Response types for plugin authors
- plugins/example-wasm: TinyGo example plugin (build with `just build-example-wasm`)

Security:
- Bump google.golang.org/grpc 1.79.2 → 1.79.3 (CVE authorization bypass fix)

All new dependencies are BSD-3/Apache-2.0/MIT licensed.

v0.1.0 — Phases 1–5 complete

Core proxy:
- OpenRouter, OpenAI, Anthropic providers
- Model-based routing with priority failover and exponential backoff
- Weighted API key selection with per-model allowlists
- Native passthrough endpoint (/native/{provider}/*)
- In-memory LRU response cache with TTL

Plugin system (built-in):
- Rate limiter (token bucket, global or per-IP)
- Request logger (structured slog)
- Prometheus metrics (OTel SDK, /metrics endpoint)
- Distributed tracing (OTel OTLP HTTP; no-op when unconfigured)

Production readiness:
- Config hot-reload (mtime polling, atomic engine state swap)
- Graceful shutdown (SIGINT/SIGTERM)
- Health check (/healthz)
- Docker image (multi-stage, distroless)
- Race-detector clean test suite

Performance (Apple M2):
- Dispatch overhead: ~1.4μs (target <50μs)
- Full HTTP round-trip overhead: ~48μs non-streaming