The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands using the Bash tool with user-supplied file paths (e.g., percli lint -f <file>). This pattern creates a potential command injection surface if the agent does not properly sanitize the filenames before passing them to the shell.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface because it processes untrusted resource definitions.
Ingestion points: Untrusted JSON and YAML resource files are ingested through the Read and Glob tools.
Boundary markers: The skill lacks explicit boundary markers or instructions to the agent to disregard any embedded instructions within the dashboard definitions.
Capability inventory: The agent has access to sensitive tools such as Bash, Write, and Edit, which could be exploited if malicious instructions in the data influence the agent's behavior during the fix-and-revalidate cycle.
Sanitization: No sanitization logic is described to filter or validate the contents of the files before processing them.

perses-lint