Docs · Prompt and RAG governance
Prompt registry and RAG source workflows
See how SentinelAI governs prompts and retrieval sources with versioning, AI-system linkage, evaluation posture, and citation-aware source records.
Overview
Prompt and RAG governance help teams manage the instructions and knowledge sources that shape AI-system behavior after deployment. SentinelAI keeps those assets in dedicated records so prompt versions, retrieval configuration, and governed source changes stay visible to reviewers.
This page is part of the public SentinelAI documentation layer. It is meant to accelerate orientation and evaluation while staying aligned with the product’s governance-focused positioning and messaging guardrails.
Prompt registry
Versioned prompt records capture prompt kind, owner, target-model context, prompt template, system prompt, variables, and test cases in one place instead of scattering them across code and tickets.
RAG source registry
Governed RAG sources track source kind, version, ingestion posture, citation labels, and the AI systems that rely on the content for grounded responses.
- Track retrieval sources such as policies, FAQs, files, URLs, runbooks, and other governed knowledge assets.
- Preserve version and ingestion posture so source changes can be reviewed alongside prompt and release changes.
- Use AI-system linkage to understand which deployed experiences depend on which knowledge assets.
Evaluation and release linkage
Prompt and retrieval governance become more durable when suite definitions, release approvals, and runtime incidents all refer back to the same governed prompt and source records.