The Agent Skills Directory

[Prompt Injection] (SAFE): No instructions found attempting to override agent behavior or bypass safety filters. Scope warnings are descriptive and appropriate for the task.
[Data Exposure & Exfiltration] (SAFE): While the skill accesses project configuration files (package.json, pom.xml, IaC files) to determine the technology stack, it lacks any network-capable tools or patterns that would facilitate data exfiltration. All operations are local and read-only.
[Indirect Prompt Injection] (LOW): The skill analyzes untrusted external data (project source code and configuration files).
Ingestion points: Files read via Read, Glob, and Grep tools.
Boundary markers: None explicitly defined in the prompt templates.
Capability inventory: Limited to read-only tools and calling other defined skills; lacks file-write or command execution capabilities.
Sanitization: None specified.
Assessment: Severity is LOW as the skill's output is intended for documentation and human review, not for driving high-privilege automated decisions.
[Unverifiable Dependencies] (SAFE): No external package installations or remote script downloads are performed.
[Obfuscation] (SAFE): No encoded content, zero-width characters, or homoglyphs were detected in the skill definitions.

gap-analysis