Glossary · 6 · Quality, risk and governance
Agent permission scope
Also known as: Agent scope, Agent permissions, Least agency
German: Berechtigungsumfang eines Agenten
The agent permission scope is the declared set of systems, data, tools and actions an AI agent may use for a given task, together with the limits on each — which records it may read, which it may change, up to which amount, in which environment and for how long.
- Intermediate
- Technical project managers
- Developers
In one sentence
Agent permission scope: what an AI agent may read, change and trigger, in which systems and for how long — the first boundary of every agent.
Example
A maintenance-planning agent may read the historian and the CMMS, create draft work orders and nothing else; it may not release work orders or touch PLC parameters.
Explained in context
This term is a station of a mini course
- When an AI agent crosses the line: detect, contain and restore operationsStation: The boundary in three parts · Scope, allowlist and own identityStart the course
Context cards connect this term with others to answer one question. Also in British English and German.
- Permission, capability, mandate: the three parts of an AI agent’s boundaryWhich boundary does an AI agent cross when it “goes too far” — and why does it matter which one?UKDeutsch
- When an AI agent crosses the line: detect, contain and restore operationsWhat has to be in place when an AI agent does something it isn’t allowed to do — in IT systems, IoT fleets, OT and on machines — and how is it detected, stopped and released again?UKDeutsch
How it applies
- Boundary in writing: The scope is the written answer to “what may this agent do?”. It names systems (ERP, ticketing, historian, device cloud), the actions per system (read, draft, write, delete, trigger) and limits (amounts, number of records, environments, time windows).
- Enforced outside the model: A scope that only lives in the system prompt is a wish. It is enforced by the accounts, access tokens and gateways behind the tools, ideally with a separate agent identity per agent.
- Least agency: OWASP's agentic guidance applies least privilege to autonomy itself: give an agent the smallest scope and the shortest-lived credentials that still let it finish the task.
- System types differ: In IT systems a scope limits records and transactions; in IoT fleets it limits which devices and which commands; in OT it should default to read-only, with writes only through the operator's HMI; on machines and robots it never includes safety functions.
Why it matters on your learning path
- Technical writers: The scope belongs in the system description and in the instructions for whoever operates the agent.
- Technical marketers: A clear scope is a selling point for cautious industrial buyers.
- Technical project managers: Make the scope part of the release decision and of every change request.
- Developers: Derive tokens and roles from the scope, not the other way round.