The Agent Skills Directory

[SAFE]: Extensive analysis across all threat categories reveals no malicious intent or security vulnerabilities. The skill logic is confined to document parsing and file system updates within a project-specific directory.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted markdown data from RFC files.
Ingestion points: Files matching RFC-*.md are read during validation and generation operations across all reference files.
Boundary markers: The instructions do not define specific delimiters for separating data from instructions within the documents.
Capability inventory: The skill is limited to file system read/write operations for documentation purposes and does not possess capabilities for network communication, subprocess execution, or shell command invocation.
Sanitization: Content is processed using structured extraction rules (metadata keys, specific terminology patterns), which restricts the agent's interpretation of document body text as commands.

platonic-specs