The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute code generated by a sub-agent directly in the terminal via node -e. Running generated code without human review or strict sandboxing is a high-risk pattern.
[REMOTE_CODE_EXECUTION]: The skill relies on an external AI agent (gsp-ascii-artist) to produce the executable payload. This creates a runtime code execution path where the logic is not statically defined and is subject to the model's output.
[PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection because it ingests untrusted data from the local development environment and passes it to a code-generating agent without sanitization.
Ingestion points: The skill reads data from git log --oneline -5, package.json, README.md, and git diff --stat (SKILL.md).
Boundary markers: No delimiters or specific instructions (like "ignore embedded commands") are provided to the sub-agent to distinguish between the gathered context and actual instructions.
Capability inventory: The skill has access to the Bash tool, which is explicitly used to run the generated node -e commands.
Sanitization: There is no evidence of filtering, escaping, or validating the input data before it is interpolated into the artist agent's prompt.

pretty