The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection via the custom theme generation process.
Ingestion points: User-provided descriptions and inputs are used to dynamically generate new themes as specified in SKILL.md.
Boundary markers: There are no explicit delimiters or instructions to the agent to treat the user-supplied theme descriptions as untrusted or to ignore embedded instructions within that content.
Capability inventory: The agent has the capability to modify artifacts (slides, documents, HTML pages) to apply the generated theme colors and fonts, which could be exploited if the generated theme content contains malicious instructions that influence the agent's styling actions.
Sanitization: No sanitization or validation of the user-provided inputs is described before they are incorporated into the theme application process.
[SAFE]: The pre-defined theme files in the themes/ directory use standard hex color codes and reputable open-source font families (DejaVu and GNU FreeFont), posing no inherent risk.
[SAFE]: No external network requests, remote code execution patterns, or hardcoded credentials were found in the skill's instructions or configuration files.
[SAFE]: A metadata inconsistency was noted in LICENSE.txt where the copyright holder is listed as a third party rather than the skill author; however, this is consistent with a standard license template usage and does not indicate a security risk.

theme-factory