The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill suggests the installation of the knap package via npm install -g knap. This is the primary tool used by the skill for rendering templates.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources and local files, which presents a surface for indirect prompt injection.
Ingestion points: Data is ingested via defuddle parse from URLs, through the --data flag for local files, and via standard input (SKILL.md).
Boundary markers: The tool uses standard template delimiters ({{, {%), though the instructions do not specify additional agent-level boundary markers for untrusted content.
Capability inventory: The skill is capable of writing files to the local file system using the -o (output) and --output-dir flags (SKILL.md).
Sanitization: The tool includes built-in filters like safe_name for generating safe filenames from data and yaml_property for correctly formatting frontmatter properties (SKILL.md).

knap