# Security & identity

Author: Murali Sid (https://linkedin.com/in/muralisid)
Source: https://www.agenticarchitectureskills.com/library/layers/r10-security-and-identity
Published: 2026-09-18
Updated: 2026-09-18
Licence: CC BY-SA 4.0, https://creativecommons.org/licenses/by-sa/4.0/
Cite as: Murali Sid. Security & identity. Agentic Architecture Skills. https://www.agenticarchitectureskills.com/library/layers/r10-security-and-identity

---


Status: complete (published 2026-08-19). Scope finalized at Batch B kickoff with maintainer POV; see findings.md and sources.md.

## Scope inventory (v0) [#scope-inventory-v0]

* **Identity**: non-human identity management as a discipline, agent identity lifecycle with Sponsor/Owner accountability, the Levels 1-3 taxonomy, delegation chains (user to agent to sub-agent) with purpose binding
* **Access**: OAuth on-behalf-of and token exchange, scoped short-lived credentials, RBAC/ABAC/ReBAC via policy engines, secrets vaulting, entitlement enforcement always outside the model
* **Threats**: prompt injection defense in depth, memory poisoning, tool supply chain vetting (MCP server provenance), sandbox escape, data exfiltration, endpoint and browser isolation for computer-use agents
* **Operations**: agent activity in SIEM/UEBA, incident response and kill switches, red teaming, audit at human-action granularity with tamper evidence

## Challenged-default candidates [#challenged-default-candidates]

Dedicated NHI security products vs extending the existing IAM estate; AI-firewall products vs layered controls at gateway and tool layer. Each track proposes its final list at kickoff.

## Files [#files]

`findings.md` and `sources.md` carry this track's research. The scoping brief and the vendor map were archived on 2026-08-23 (see `archive/`, decision D038).