context-engineering — agentic threat model

Not certain from the listing — The specific foundation model is not defined, but the skill's reliance on rules files makes it highly susceptible to indirect prompt injection and adversarial reprogramming if malicious instructions are introduced into the project context.

The skill directly structures project context and configures rules files. This introduces a risk of data poisoning or context manipulation if an attacker can write to or influence the project files that the skill reads.

As an orchestration skill, it manages task switches and context structuring. Vulnerabilities in the host framework could allow malicious rules files to execute unauthorized tool calls or bypass system prompts.

Not certain from the listing — The deployment environment is unspecified, though as an open-source developer skill, it likely runs locally, meaning compromised rules files could potentially access local file systems depending on the host sandbox.

Not certain from the listing — The skill claims to manage task switches when output degrades, implying some form of quality monitoring, but the lack of explicit security guardrails or logging details leaves potential blind spots.

Not certain from the listing — There are no mentioned security controls, authentication mechanisms, or compliance alignments (like NIST or ISO) within this open-source skill.

The skill is designed to configure and reshape another agent's context. This creates a multi-agent trust boundary where a compromise in the context-engineering skill directly compromises the target agent's behavior.