The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill possesses a high-risk capability tier combined with an external ingestion surface.
Ingestion points: User-provided diagram descriptions or external content mentioned in requests (SKILL.md).
Boundary markers: Absent. There are no instructions to delimit or ignore instructions within the data being processed.
Capability inventory: The frontmatter explicitly requests Bash, Write, and Edit tools (SKILL.md).
Sanitization: None. The lack of logic to sanitize inputs means an attacker could embed shell commands or malicious file modifications inside a 'diagram' request that the agent might execute.
[No Code] (LOW): The skill contains metadata and descriptions but lacks any implementation scripts. Requesting Bash and Write privileges without providing specific scripts to perform those tasks suggests a reliance on the agent's raw capabilities, which bypasses the safety controls usually found in well-defined skill scripts.

ascii-art-diagram-creator