The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to audit and modify content within user-controlled directories (wiki/ and sources/). An attacker could place malicious instructions inside these files that the agent might follow during the linting process.
Ingestion points: Processes all files located in the wiki/ and sources/ folders.
Boundary markers: No explicit delimiters or instructions are provided to the agent to distinguish between wiki data and its own system instructions.
Capability inventory: The skill is authorized to read and write files (including the wiki/ directory and .llm-wiki/lint-result.yaml) and execute the llm-wiki command-line tool.
Sanitization: The instructions do not define any sanitization or filtering logic for the content being processed before it is evaluated by the model.

lint