The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process external, potentially untrusted documentation files. This creates a surface for indirect prompt injection if the files being verified contain malicious instructions intended to mislead the AI agent during the validation process.
Ingestion points: Markdown files identified in the workspace and processed via the scripts/doc-link-check.py tool or manual review as specified in SKILL.md.
Boundary markers: The skill does not currently define specific delimiters or guardrails to separate documentation content from instructions.
Capability inventory: The skill utilizes local file system access and is instructed to validate or execute commands and snippets found within documentation files.
Sanitization: The risk is mitigated by explicit instructions in references/validation.md to execute code examples within isolated environments.
[SAFE]: The provided script scripts/doc-link-check.py is a benign utility that uses standard Python libraries to resolve relative paths and verify file existence. It does not perform network operations, access sensitive system directories, or execute arbitrary code.

documentation-verification