Inferra

Inferra lets an assistant run auditable rule-based decisions through a Rete engine, with LLM-generated explanations of why each decision was reached.

What you can do

Inferra combines a Rete rule engine with retrieval-augmented generation: the rule engine produces the decision, and a language model explains the reasoning behind it. The registry description frames it as "Auditable rule-based decisions with RAG explanations: a Rete engine decides, an LLM explains why." The manifest does not list individual tools, so the exact callable operations are not documented here.

When to use it

Consider it when you need decisions that follow explicit, inspectable rules but also want a readable explanation of each outcome, for example in policy checks, eligibility screening, or other rule-driven workflows where auditability matters.

Requirements

Runs locally as a stdio process, launched with uvx from the PyPI package inferra-mcp, so a Python/uv runtime is needed. It requires the secret INFERRA_API_KEY, which the manifest says is created at ai-rete-rag.com under Settings, and an optional INFERRA_API_URL that defaults to Network access to that service is implied.