The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to manage presentation files. It uses find to locate configuration files and uvx to run a specialized slides tool for inspection and editing. These operations are restricted to the local file system and align with the skill's documented purpose.
[EXTERNAL_DOWNLOADS]: The skill dynamically installs and runs the agent-slides package using uvx. This involves downloading code from an external package registry. This behavior is standard for the requested environment (Python 3.12+ and uv) and is necessary for the skill's functionality.
[PROMPT_INJECTION]: The skill processes untrusted input from external slide files, which presents a surface for indirect prompt injection.
Ingestion points: Presentation content is read from output.pptx and slides.json (SKILL.md).
Boundary markers: There are no explicit markers or instructions to isolate the slide content from the agent's core instructions.
Capability inventory: The skill can execute commands and modify files using the uvx slides edit and uvx slides apply commands (SKILL.md).
Sanitization: No data sanitization or escaping of the ingested slide text is performed before the critique process.

slides-critique