No Excessive CAP is a governance frame for agentic AI that focuses on Capabilities, Autonomy, and Permissions. It is useful because it separates what an agent may reach, how independently it may act, and how much room it has to expand beyond intended behaviour.
Expanded Definition
No Excessive CAP is a governance lens for agentic AI that evaluates whether an agent’s Capabilities, Autonomy, and Permissions are tightly bounded to the task it is meant to perform. The term is not a formal NIST control or an OWASP standard, and usage in the industry is still evolving, so teams should treat it as an operational model rather than a fixed compliance label.
Capability asks what tools, data sources, and actions the agent can reach. Autonomy asks when it can decide and execute without human intervention. Permissions ask what the environment will actually allow it to do. That separation matters because an agent can be highly capable but narrowly permitted, or lightly capable but dangerously autonomous. A sound implementation maps those three dimensions independently instead of assuming one proxy, such as a prompt policy, is enough. For adjacent guidance on bounding identity power and privilege, the NIST SP 800-53 Rev 5 Security and Privacy Controls remains useful for privilege discipline, while the Ultimate Guide to NHIs provides the NHI governance context that makes excessive capability easier to spot.
The most common misapplication is treating a narrow prompt instruction as proof that an agent cannot exceed its intended scope, which occurs when tool access and backend permissions remain broader than the prompt.
Examples and Use Cases
Implementing No Excessive CAP rigorously often introduces workflow friction, requiring organisations to balance agent productivity against tighter approval gates, narrower tool access, and more frequent policy reviews.
- An internal support agent can read a knowledge base but cannot open tickets, change records, or message customers unless a human grants a bounded action token for that session.
- A finance agent may draft payment instructions, yet it cannot submit or release transfers because its permissions stop short of execution, aligning with least-privilege principles in NIST SP 800-53 Rev 5 Security and Privacy Controls.
- A software delivery agent can propose code changes and open pull requests, but it cannot merge to production unless a reviewer approves the final state.
- A procurement agent is allowed to compare vendors and generate quotes, yet it cannot create purchase orders or expand its own access to contract repositories.
- NHIMG’s Ultimate Guide to NHIs is useful when mapping these examples to service-account and API-key governance, because the same privilege drift patterns often appear in both NHIs and agents.
Why It Matters in NHI Security
No Excessive CAP matters because excessive capability in an agent becomes an NHI risk as soon as the agent is allowed to operate with real credentials, real data, and real side effects. When autonomy and permissions are not separated, the result is often privilege creep, hidden escalation paths, and tool misuse that security teams only discover after an incident.
This is especially important in environments where NHIs already carry excessive privilege. NHI Mgmt Group reports that 97% of NHIs carry excessive privileges, which shows how quickly unmanaged access can become the default state. In practice, No Excessive CAP is reinforced by controls in NIST SP 800-53 Rev 5 Security and Privacy Controls and by policy patterns that limit agent reach before damage is possible. The concept is also closely aligned with the governance lessons in the Ultimate Guide to NHIs, especially around visibility, rotation, and offboarding of machine identities.
Organisations typically encounter this problem after an agent has already accessed the wrong system, exfiltrated the wrong data, or triggered an unauthorised action, at which point No Excessive CAP becomes operationally unavoidable to address.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Agentic AI Top 10 and OWASP Non-Human Identity Top 10 address the attack and risk surface, while NIST CSF 2.0, NIST AI RMF and NIST Zero Trust (SP 800-207) set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| OWASP Agentic AI Top 10 | Agentic AI guidance centers on limiting tools, autonomy, and harmful action scope. | |
| OWASP Non-Human Identity Top 10 | NHI-02 | Excessive capability often manifests as overprivileged non-human identities and secret misuse. |
| NIST CSF 2.0 | PR.AC-4 | Access permissions should follow least-privilege and managed authorization principles. |
| NIST AI RMF | AI risk management requires bounding autonomy, oversight, and system impact. | |
| NIST Zero Trust (SP 800-207) | Zero trust limits implicit trust and verifies every action and access request. |
Constrain agent tools, execution rights, and escalation paths to the minimum needed for the task.
Related resources from NHI Mgmt Group
Deepen Your Knowledge
Reviewed and updated by the NHIMG editorial team on July 28, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org
