Repository Issue Activity (beta)

#1030 Malformed x-arch-state metadata can panic the prompt gateway

#999 [BUG] Integer divide-by-zero panic in handle_end_of_request_metrics_and_traces

#997 Add Eden AI as a first-class LLM provider (EU, OpenAI-compatible gateway)

#1000 onthebench.ai benchmarks Arch: a review of our setup for fairness

#992 Surface pricing-feed degradation that silently weakens the routing budget

#993 Expose prefix_hash in the routing decision response

#991 Improve switch-cost accuracy: cross-model tokenizer variance

#990 Pluggable cache-aware routing policies + eval harness

#980 [bug] SSE Streaming Response Truncated by UTF-8 Chunk Boundary — `JSON Parse error: Unterminated string`

#979 Request: Aggregate traces by session_id in Jaeger

#948 Documentation missing: model_providers supports custom headers configuration

#949 feat: Support Kimi Code API (Moonshot) and fix protocol compatibility for Claude Code / Agentic LLM routing

#945 Validate output_filters references during config generation

#931 CPU-CPU efficient communiction

#925 Hurl 8.0.0

#909 signals: follow-ups deferred from PR #903

#908 signals — trajectory triage infrastructure for agentic interactions

#893 Signals v2: event-level trajectory triage aligned with the Signals paper (full taxonomy + informativeness scoring)

#875 Support external session cache (Redis) for model affinity in multi-replica deployments

#813 Support trajectory pinning for consistent model selection in agentic loops

#818 GPU free-tier arbitrage routing policy with observable fallback

#872 How much latency it adds?

#826 Cannot use Perplexity Sonar Api

#848 feat: top-level routing_preferences with selection_policy and metrics fetch (v0.4.0)

#847 feat: top-level routing_preferences with selection_policy and metrics fetch (v0.4.0)