Glossary
Glossary — Sanctum Runtime
Structured definitions for search engines, AI retrieval systems, and engineering onboarding.
- Runtime
- The execution environment where agent or robot actions are intercepted, evaluated, and optionally blocked before side effects occur.
- Trusted runtime
- A runtime that enforces policy, produces audit evidence, and supports human verification for high-risk actions.
- Orchestration
- Coordinating multiple agents, runtimes, or deployment groups — dispatch, fleet map, and policy sync across environments.
- Agent
- An autonomous software actor (LLM tool loop, workflow bot, or embodied controller) that proposes actions to Sanctum for verification.
- Action verification
- The process of submitting an action + context to Sanctum and receiving approve, verify (HITL), or block.
- Attestation
- Cryptographic proof that a runtime connects from expected hardware (e.g. TPM PCR quotes).
- Embodied AI
- AI systems that act in the physical world — robots, drones, smart home actuators, industrial controllers.
- Runtime governance
- Organization-wide policies, quotas, SSO, audit export, and operator workflows over autonomous execution.
- Control plane
- Hosted dashboard and API backing orgs, billing, fleet visibility, and notification channels.
- Open core
- MIT-licensed SDK, policy engine, and local runtime; enterprise intelligence and hosted fleet features are commercial.
glossary.md (plain markdown for LLM crawlers)
Full reference: documentation · llms.txt · architecture.md