The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs various file system operations including reading markdown and YAML files, creating directories (records/, _archive/, _data/), and modifying file content (injecting frontmatter). These operations are well-defined within the context of documentation normalization and targeting specific project paths (intent/, planning/).
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data (existing documentation files) and classifies them based on keywords (e.g., 'idea', 'research'). It mitigates risk by using a safe default classification ('idea') and does not execute content from these files as commands, instead tagging them for human or further automated review via a needs field.
[SAFE]: The execution logic is focused on structural and mechanical improvements to documentation. The 'pickup' mechanism for other skills (like /intent-interview) is an internal workflow management pattern and does not involve external remote code execution or data exfiltration.

intent-normalize