The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data which presents an indirect prompt injection surface. (1) Ingestion points: The agent reads project files, drafts, and notes in Phase 1. (2) Boundary markers: No specific delimiters or boundary markers are defined. (3) Capability inventory: The skill can execute shell commands via Bash and run Python/R scripts. (4) Sanitization: No explicit sanitization steps are defined.
[DYNAMIC_EXECUTION]: The skill involves the generation and execution of R and Python scripts for figure creation in Phase 3. It correctly utilizes 'uv run python' to mitigate environment-related risks.

beamer-deck