The Agent Skills Directory

[Prompt Injection] (LOW): The skill possesses an attack surface for Indirect Prompt Injection. It is designed to ingest and aggregate data from multiple YAML files (character profiles, room environments, costume details) to synthesize prompts for image generation.
Ingestion points: PHOTO-SET-8.yml specifically gathers context from player.yml, costume.yml, and ROOM.yml via the Gather Context step.
Boundary markers: There are no explicit delimiters or 'ignore instructions' warnings used when interpolating these variables into the final prompt templates.
Capability inventory: The skill (via visualize.py) possesses the capability to send synthesized prompts to external image generation providers.
Sanitization: The provided configuration files do not define any sanitization, escaping, or validation logic for the external data being processed.
[Data Exfiltration] (LOW): The skill documentation (GLANCE.yml, skill-snitch-report.md) confirms network operations to external image generation providers, including OpenAI, Google, Stability, and Replicate. These domains are not on the approved whitelist for the exfiltration analysis scope. However, these operations are central to the skill's purpose and are not associated with sensitive file access (e.g., SSH keys or credentials).

visualizer