The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data such as notes, drafts, and external links, creating an indirect prompt injection surface. * Ingestion points: User material (notes, fragments, draft files, links, feedback) as specified in SKILL.md. * Boundary markers: The instructions do not define specific delimiters or warnings for the ingested user content. * Capability inventory: The skill is limited to text analysis and generation across all files in the steps/ directory, along with potential file-write suggestions in SKILL.md. * Sanitization: No explicit sanitization of potentially malicious instructions embedded in user material or external links is performed.
[SAFE]: The skill exhibits no signs of malicious intent. There are no hardcoded credentials, obfuscated commands, persistence mechanisms, or unauthorized network requests. All file operations are scoped to the user's project directory for the purpose of maintaining editorial consistency.

essay