duende-skills — agentic threat model
The duende-skills agent is a specialized plugin for Claude Code focused on identity and authentication architecture. Its primary risk lies in generating or reviewing security-critical code (OAuth/OIDC, BFF patterns) where subtle vulnerabilities or malicious code generation could lead to downstream system compromises.
OWASP AIVSS score rationale
| Autonomy of Action | 0.30 | |
| Goal-Driven Planning | 0.40 | |
| Self-Modification | 0.10 | |
| Dynamic Tool Use | 0.30 | |
| Persistent Memory | 0.20 | |
| Contextual Awareness | 0.60 | |
| Dynamic Identity | 0.20 | |
| Multi-Agent Interactions | 0.50 | |
| Non-Determinism | 0.50 | |
| Opacity & Reflexivity | 0.40 |
Scored with the canonical OWASP AIVSS formula (AIVSS calculator reference); agentic risk factors estimated from the agent’s described capabilities.
MAESTRO 7-layer threat model
Per-layer threats for this agent. Layers tagged “not certain from listing” are general, caveated commentary where the public description didn’t pin that layer.
Not certain from the listing — relies on Claude Code's underlying foundation models (Anthropic Claude). Threats include prompt injection that could trick the model into generating insecure OAuth/OIDC configurations or weak cryptographic implementations.
Not certain from the listing — likely utilizes a local or remote knowledge base of IdentityServer, ASP.NET Core, and OAuth specifications. Poisoning this reference data could lead the agent to recommend deprecated or vulnerable security patterns.
Operates as a plugin within the Claude Code agent framework. It exposes specialized skills and subagents. Vulnerabilities could arise if the orchestration layer allows these subagents to execute unvalidated code or write directly to sensitive local configuration files.
Not certain from the listing — runs locally within the user's development environment via Claude Code. If the host environment lacks sandboxing, a compromised plugin could read local secrets, developer certificates, or ASP.NET Core appsettings.json files.
Not certain from the listing — relies on Claude Code's built-in logging and execution guardrails. There is a risk of silent failures or 'hallucinated' security advice that passes basic syntax checks but introduces logical authentication bypasses.
Designed specifically to assist with security and compliance (OAuth/OIDC, BFF patterns). However, the plugin itself does not explicitly list independent compliance certifications (e.g., SOC2) or automated policy enforcement mechanisms within the listing.
Explicitly surfaces subagents to generate and review secure identity code. This multi-agent interaction within Claude Code introduces risks of cascading trust, where a compromised subagent could pass flawed security code to another subagent for validation, bypassing human review.
MAESTRO — the 7-layer agentic threat-modeling framework (Cloud Security Alliance / Ken Huang).