Llm Router Harness¶
Component id: llm_router
Source contract: BlockContract for llm_router.
Overview¶
LLM routing layer via OpenRouter and Ollama with resource enforcement, model discovery, hardware-aware recommendations, and LiteLLM-based completion with custom callback hooks.
When to use:
- Routing LLM calls through OpenRouter or local Ollama backends
- Local inference with hardware-aware model selection
- Resource-bounded LLM usage with token tracking
Good at¶
- open-ended generation
- structured-output prompts with explicit format
- summarisation and rewriting
Not suited for¶
- deterministic arithmetic
- verifiable factual lookup without retrieval grounding
- long-horizon planning without intermediate verification
Inputs and outputs¶
- Required:
prompt - Optional:
model,system_prompt,max_tokens,temperature - Outputs:
completion_text
Mitigates¶
Known Failure Modes¶
hallucination: Fluent text not grounded in any source.timeout: Provider exceeded the call timeout.rate_limit: Provider returned 429 or quota-exceeded.
State Surface¶
- Persistence:
session - Queryable via:
scorecard - Retention:
90d
Extension Points¶
training_signal_inputs:('agent_routing_outcomes', 'provider_errors')validation_gate:unit_plus_codex_reviewallowed_mutation_surfaces:('routing_policy', 'provider_thresholds')
Verification¶
- Method:
external_evidence_or_human_review - Human review gate:
False - Fallback block:
none