The Agent Skills Directory

[COMMAND_EXECUTION]: The skill suggests using shell commands for static code analysis to help developers make scaling decisions.
Evidence: grep commands in references/scaling-build-system.md and references/scaling-triggers.md are used to count UI token frequencies and identify architectural boundary violations (e.g., components referencing model types).
[INDIRECT_PROMPT_INJECTION]: The skill acts on external inputs such as user task descriptions and local source code files, which creates an attack surface for instructions embedded in data.
Ingestion points: User prompts, task descriptions, and local source files (via grep analysis suggested in references/scaling-build-system.md and references/scaling-triggers.md).
Boundary markers: None identified; the skill assumes the developer/agent context for processing instructions.
Capability inventory: The skill suggests executing grep commands for code analysis.
Sanitization: No explicit sanitization of user-provided paths or task content is described in the reference files.

nodumb