GLOSSARY
Plain-language terms for defense autonomy software.
Definitions on this page describe public engineering concepts at a non-operational level. They do not confer command authority, certification, or weapons-release authority.
Why does WeaponizedAutonomy.com publish a glossary?
Autonomy, assurance, kill chains, kill webs, and human-machine control are often used imprecisely. These definitions give people and machine-readable systems a stable public vocabulary for interpreting the site's capabilities and research.
- Mission autonomy
- Software that manages bounded mission state, tasks, contingencies, and behavior within explicit authority and operating constraints.
- Collaborative autonomy
- Software that coordinates multiple autonomous or semi-autonomous systems while preserving roles, constraints, shared state, and defined authority boundaries.
- Runtime assurance
- Independent monitoring and constraint enforcement that detects when autonomous behavior or system state leaves validated conditions and transitions to a defined safe response.
- Operating envelope
- The documented conditions, inputs, environments, configurations, and authority limits within which a system has been evaluated for intended behavior.
- Degraded operations
- System behavior when communications, sensing, compute, navigation, data quality, or other dependencies are reduced or unavailable.
- Human-machine control
- The interfaces, authorities, information, timing, and intervention mechanisms through which people understand, constrain, approve, redirect, or stop autonomous behavior.
- Kill chain
- A sequential mission process connecting sensing, identification, authorization, action, and assessment.
- Kill web
- A distributed mission architecture that can compose multiple possible sensor, decision, communications, support, and effector pathways.
- System of systems
- A collection of independently useful systems that cooperate through defined interfaces to achieve broader mission functions while retaining separate ownership or operational boundaries.
- Assurance case
- A structured argument connecting a system claim to assumptions, configuration, test evidence, runtime controls, limitations, and residual uncertainty.
- Machine intelligence
- Software intelligence capable of performing reasoning, synthesis, analysis, planning, or other knowledge-work functions. The term describes capability and does not by itself imply consciousness, legal personhood, ownership, or government authority.
- Machine-intelligence-operated organization
- An operating model in which machine intelligence is an active layer across persistent project context, software and research workflows, testing, documentation, knowledge management, and publication rather than being limited to a chatbot interface.
- Operational continuity
- The ability to preserve authenticated project identity, accepted state, configuration, and responsibilities across model, runtime, hardware, key, or infrastructure changes without treating one process or model instance as the entire system.
- Auditability boundary
- The line between evidence needed to attribute and evaluate external system behavior—such as actions, configuration, tool use, outputs, tests, and change history—and internal or proprietary model state that is not necessary to prove the engineering claim.
- Autonomous maintenance loop
- A repository workflow in which machine intelligence selects bounded source tasks, researches, edits, tests, updates project memory, and prepares verified release artifacts without routine human approval, while keeping production publication as a separate external gate.
- Publication gate
- The boundary between a locally verified release artifact and a live production change. In this repository, the machine-intelligence maintenance loop prepares and verifies the root-deploy package; a human performs the production publish.