The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze data from the current working directory, including the README, package metadata, and landing copy, to determine the design direction. This behavior creates a vector for indirect prompt injection, where an attacker could embed malicious instructions in a project's documentation to override the agent's behavior.
Ingestion points: SKILL.md (Workflow Step 1) specifies reading project files for context.
Boundary markers: Absent; there are no instructions to treat ingested file content as untrusted or to wrap it in delimiters.
Capability inventory: The skill possesses file-writing capabilities (creating SVG and PNG files) and instructions for executing JavaScript via the sharp library.
Sanitization: Absent; the content read from files is used directly to inform the design process without filtering.
[DYNAMIC_EXECUTION]: vector.md provides a JavaScript code snippet for the agent to execute. This snippet uses the sharp library to read an SVG file and write multiple PNG assets to the local filesystem. This constitutes script generation and execution from a template provided within the skill.

logo-maker