The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local scripts and CLI tools as part of its core functionality, including skill-auditor, converter, heal-skill, and the ao tool. These operations are restricted to the local environment and are necessary for the skill's purpose of building and auditing other skills.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface when processing external skill files or using the from-pattern mode.
Ingestion points: scripts/init.sh (via the --absorb flag) and scripts/build.sh (via the from-pattern mode).
Boundary markers: Currently, the skill does not wrap processed content in explicit security boundary markers.
Capability inventory: The skill can create files and execute local validation scripts.
Sanitization: Basic validation is performed on skill names to ensure they follow a lowercase-hyphen slug format.
[EXTERNAL_DOWNLOADS]: The documentation references absorbing content from official repositories associated with Anthropics. These references are treated as trusted sources for development templates and do not represent a security risk.

skill-builder